@wlearn/automl 0.1.0 → 0.2.0

package/dist/automl.mjs

@@ -0,0 +1,2 @@
+var __automl=(()=>{var Jn=(s=>typeof require<"u"?require:typeof Proxy<"u"?new Proxy(s,{get:(t,n)=>(typeof require<"u"?require:t)[n]}):s)(function(s){if(typeof require<"u")return require.apply(this,arguments);throw Error('Dynamic require of "'+s+'" is not supported')});var S=(s,t)=>()=>(t||s((t={exports:{}}).exports,t),t.exports);var z=S((oc,ve)=>{var $=class extends Error{constructor(t,n){super(t),this.name="WlearnError",this.code=n||"ERR_WLEARN"}},St=class extends ${constructor(t){super(t,"ERR_BUNDLE"),this.name="BundleError"}},_t=class extends ${constructor(t){super(t,"ERR_REGISTRY"),this.name="RegistryError"}},vt=class extends ${constructor(t){super(t,"ERR_VALIDATION"),this.name="ValidationError"}},Ft=class extends ${constructor(t="Model is not fitted. Call fit() first."){super(t,"ERR_NOT_FITTED"),this.name="NotFittedError"}},At=class extends ${constructor(t="Model has been disposed."){super(t,"ERR_DISPOSED"),this.name="DisposedError"}};ve.exports={WlearnError:$,BundleError:St,RegistryError:_t,ValidationError:vt,NotFittedError:Ft,DisposedError:At}});var G=S((ac,Fe)=>{var{ValidationError:N}=z();function Kn(s,t="auto"){if(s&&typeof s=="object"&&!Array.isArray(s)&&s.data){let{data:n,rows:e,cols:i}=s;if(!(n instanceof Float64Array)){if(t==="error")throw new N("Expected Float64Array in typed matrix");return{data:new Float64Array(n),rows:e,cols:i}}return{data:n,rows:e,cols:i}}if(Array.isArray(s)&&Array.isArray(s[0])){if(t==="error")throw new N('Input coercion disabled (coerce: "error"). Pass { data: Float64Array, rows, cols } instead of number[][].');let n=s.length,e=s[0].length,i=new Float64Array(n*e);for(let r=0;r<n;r++)for(let o=0;o<e;o++)i[r*e+o]=s[r][o];if(t==="warn"){let r=i.byteLength;console.warn(`@wlearn/core: Converted number[][] to Float64Array (copied ${(r/1024).toFixed(1)} KB, shape ${n}x${e}). For performance, pass { data, rows, cols }.`)}return{data:i,rows:n,cols:e}}throw new N("X must be number[][] or { data: Float64Array, rows, cols }")}function Hn(s){return s instanceof Int32Array||s instanceof Float32Array||s instanceof Float64Array?s:new Float64Array(s)}function Zn(s,t,n){if(!t||!n||t<1||n<1)throw new N(`Invalid dimensions: rows=${t}, cols=${n}`);if(!(s instanceof Float32Array)&&!(s instanceof Float64Array))throw new N("data must be Float32Array or Float64Array");if(s.length!==t*n)throw new N(`data.length (${s.length}) !== rows * cols (${t*n})`);return{data:s,rows:t,cols:n}}function Qn(s){if(!s||typeof s!="object")throw new N("Matrix must be an object");let{data:t,rows:n,cols:e}=s;if(typeof n!="number"||typeof e!="number"||n<1||e<1)throw new N(`Invalid dimensions: rows=${n}, cols=${e}`);if(!(t instanceof Float32Array)&&!(t instanceof Float64Array))throw new N("data must be Float32Array or Float64Array");if(t.length!==n*e)throw new N(`data.length (${t.length}) !== rows * cols (${n*e})`);return s}Fe.exports={normalizeX:Kn,normalizeY:Hn,makeDense:Zn,validateMatrix:Qn}});var kt=S((cc,Ae)=>{var tr=new Uint32Array([1116352408,1899447441,3049323471,3921009573,961987163,1508970993,2453635748,2870763221,3624381080,310598401,607225278,1426881987,1925078388,2162078206,2614888103,3248222580,3835390401,4022224774,264347078,604807628,770255983,1249150122,1555081692,1996064986,2554220882,2821834349,2952996808,3210313671,3336571891,3584528711,113926993,338241895,666307205,773529912,1294757372,1396182291,1695183700,1986661051,2177026350,2456956037,2730485921,2820302411,3259730800,3345764771,3516065817,3600352804,4094571909,275423344,430227734,506948616,659060556,883997877,958139571,1322822218,1537002063,1747873779,1955562222,2024104815,2227730452,2361852424,2428436474,2756734187,3204031479,3329325298]),R=new Uint32Array(64);function L(s,t){return s>>>t|s<<32-t}function er(s,t,n){for(let f=0;f<16;f++)R[f]=(t[n]<<24|t[n+1]<<16|t[n+2]<<8|t[n+3])>>>0,n+=4;for(let f=16;f<64;f++){let u=L(R[f-15],7)^L(R[f-15],18)^R[f-15]>>>3,d=L(R[f-2],17)^L(R[f-2],19)^R[f-2]>>>10;R[f]=R[f-16]+u+R[f-7]+d>>>0}let e=s[0],i=s[1],r=s[2],o=s[3],a=s[4],l=s[5],h=s[6],c=s[7];for(let f=0;f<64;f++){let u=L(a,6)^L(a,11)^L(a,25),d=a&l^~a&h,m=c+u+d+tr[f]+R[f]>>>0,p=L(e,2)^L(e,13)^L(e,22),g=e&i^e&r^i&r,w=p+g>>>0;c=h,h=l,l=a,a=o+m>>>0,o=r,r=i,i=e,e=m+w>>>0}s[0]=s[0]+e>>>0,s[1]=s[1]+i>>>0,s[2]=s[2]+r>>>0,s[3]=s[3]+o>>>0,s[4]=s[4]+a>>>0,s[5]=s[5]+l>>>0,s[6]=s[6]+h>>>0,s[7]=s[7]+c>>>0}function sr(s){let t=s instanceof Uint8Array?s:new Uint8Array(s),n=t.length,e=n*8,i=n+9+63&-64,r=new Uint8Array(i);r.set(t),r[n]=128;let o=new DataView(r.buffer);o.setUint32(i-4,e>>>0,!1),e>4294967295&&o.setUint32(i-8,e/4294967296>>>0,!1);let a=new Uint32Array([1779033703,3144134277,1013904242,2773480762,1359893119,2600822924,528734635,1541459225]);for(let h=0;h<i;h+=64)er(a,r,h);let l="";for(let h=0;h<8;h++)l+=a[h].toString(16).padStart(8,"0");return l}Ae.exports={sha256Sync:sr}});var qe=S((lc,ke)=>{var nr=new Uint8Array([87,76,82,78]),rr=1,ir=16,or={FLOAT32:"float32",FLOAT64:"float64",INT32:"int32"};ke.exports={BUNDLE_MAGIC:nr,BUNDLE_VERSION:rr,HEADER_SIZE:ir,DTYPE:or}});var tt=S((fc,Re)=>{var{BUNDLE_MAGIC:Ee,BUNDLE_VERSION:at,HEADER_SIZE:E}=qe(),{BundleError:C}=z(),{sha256Sync:Ie}=kt();function Q(s){return s==null||typeof s!="object"?JSON.stringify(s):Array.isArray(s)?"["+s.map(n=>Q(n)).join(",")+"]":"{"+Object.keys(s).sort().map(n=>JSON.stringify(n)+":"+Q(s[n])).join(",")+"}"}var qt=new TextEncoder,Et=new TextDecoder;function ar(s,t){let n=[...t].sort((d,m)=>d.id<m.id?-1:d.id>m.id?1:0),e=0,i=[],r=[];for(let d of n){let m=d.data instanceof Uint8Array?d.data:new Uint8Array(d.data),p=Ie(m),g={id:d.id,offset:e,length:m.length,sha256:p};d.mediaType&&(g.mediaType=d.mediaType),i.push(g),r.push(m),e+=m.length}let o={...s,bundleVersion:at},a=qt.encode(Q(o)),l=qt.encode(Q(i)),h=E+a.length+l.length+e,c=new Uint8Array(h),f=new DataView(c.buffer);c.set(Ee,0),f.setUint32(4,at,!0),f.setUint32(8,a.length,!0),f.setUint32(12,l.length,!0),c.set(a,E),c.set(l,E+a.length);let u=E+a.length+l.length;for(let d of r)c.set(d,u),u+=d.length;return c}function Pe(s){let t=s instanceof Uint8Array?s:new Uint8Array(s);if(t.length<E)throw new C(`Bundle too small: ${t.length} bytes (minimum ${E})`);for(let f=0;f<4;f++)if(t[f]!==Ee[f])throw new C("Invalid bundle magic (expected WLRN)");let n=new DataView(t.buffer,t.byteOffset,t.byteLength),e=n.getUint32(4,!0);if(e!==at)throw new C(`Unsupported bundle version: ${e} (expected ${at})`);let i=n.getUint32(8,!0),r=n.getUint32(12,!0);if(E+i+r>t.length)throw new C(`Bundle truncated: header declares ${E+i+r} bytes but got ${t.length}`);let o,a;try{o=JSON.parse(Et.decode(t.subarray(E,E+i)))}catch(f){throw new C(`Invalid manifest JSON: ${f.message}`)}try{a=JSON.parse(Et.decode(t.subarray(E+i,E+i+r)))}catch(f){throw new C(`Invalid TOC JSON: ${f.message}`)}let l=E+i+r,h=t.length-l;for(let f=0;f<a.length;f++){let u=a[f];if(u.offset<0||u.length<0||u.offset+u.length>h)throw new C(`TOC entry "${u.id}" out of bounds: offset=${u.offset}, length=${u.length}, blobRegion=${h}`);for(let d=f+1;d<a.length;d++){let m=a[d],p=u.offset,g=u.offset+u.length,w=m.offset,b=m.offset+m.length;if(p<b&&w<g&&u.length>0&&m.length>0)throw new C(`TOC entries "${u.id}" and "${m.id}" overlap`)}}let c=t.subarray(l);return{manifest:o,toc:a,blobs:c}}function cr(s){return qt.encode(Q(s))}function lr(s){let t=s instanceof Uint8Array?s:new Uint8Array(s);return JSON.parse(Et.decode(t))}function fr(s){let{manifest:t,toc:n,blobs:e}=Pe(s);for(let i of n){let r=e.subarray(i.offset,i.offset+i.length),o=Ie(r);if(o!==i.sha256)throw new C(`SHA-256 mismatch for "${i.id}": expected ${i.sha256}, got ${o}`)}return{manifest:t,toc:n,blobs:e}}Re.exports={encodeBundle:ar,decodeBundle:Pe,encodeJSON:cr,decodeJSON:lr,validateBundle:fr}});var ct=S((hc,ze)=>{var{RegistryError:V}=z(),{decodeBundle:Me}=tt(),W=new Map;function hr(s,t){if(typeof s!="string"||!s.includes("@"))throw new V(`Invalid typeId "${s}": must contain "@" (e.g. "wlearn.liblinear.classifier@1")`);if(typeof t!="function")throw new V("loaderFn must be a function");W.set(s,t)}async function ur(s){let{manifest:t,toc:n,blobs:e}=Me(s),{typeId:i}=t;if(!i)throw new V("Bundle manifest missing typeId");let r=W.get(i);if(!r){let o=[...W.keys()],a=o.length>0?`Registered loaders: ${o.join(", ")}`:"No loaders registered";throw new V(`No loader registered for typeId "${i}". ${a}. Install the corresponding @wlearn/* package and import it to register the loader.`)}return await r(t,n,e)}function dr(s){let{manifest:t,toc:n,blobs:e}=Me(s),{typeId:i}=t;if(!i)throw new V("Bundle manifest missing typeId");let r=W.get(i);if(!r){let a=[...W.keys()],l=a.length>0?`Registered loaders: ${a.join(", ")}`:"No loaders registered";throw new V(`No loader registered for typeId "${i}". ${l}. Install the corresponding @wlearn/* package and import it to register the loader.`)}let o=r(t,n,e);if(o&&typeof o.then=="function")throw new V(`Loader for "${i}" returned a Promise. Use async load() instead of loadSync().`);return o}function mr(){return new Map(W)}ze.exports={register:hr,load:ur,loadSync:dr,getRegistry:mr}});var Pt=S((uc,Ne)=>{var It=class{#t;#e;constructor(t,n){this.#t=t,this.#e=n}get name(){return this.#t}get estimator(){return this.#e}get isFitted(){return this.#e.isFitted}get isTransformer(){return typeof this.#e.transform=="function"}};Ne.exports={Step:It}});var je=S((dc,Ce)=>{var{Step:pr}=Pt(),{DisposedError:gr,NotFittedError:br,ValidationError:lt}=z(),{encodeBundle:wr,decodeBundle:yr}=tt(),{register:xr,load:Le}=ct(),Be="wlearn.pipeline@1",Oe=!1,ft=class s{#t;#e=!1;#s=!1;constructor(t){if(this.#t=t.map(([n,e])=>new pr(n,e)),this.#t.length===0)throw new lt("Pipeline requires at least one step")}#n(){if(this.#s)throw new gr("Pipeline has been disposed.")}#i(){if(this.#n(),!this.#e)throw new br("Pipeline is not fitted. Call fit() first.")}#r(t){let n=t;for(let e=0;e<this.#t.length-1;e++)n=this.#t[e].estimator.transform(n);return n}fit(t,n){this.#n();let e=t;for(let r=0;r<this.#t.length-1;r++){let o=this.#t[r].estimator;typeof o.fitTransform=="function"?e=o.fitTransform(e,n):(o.fit(e,n),e=o.transform(e))}return this.#t[this.#t.length-1].estimator.fit(e,n),this.#e=!0,this}predict(t){this.#i();let n=this.#r(t);return this.#t[this.#t.length-1].estimator.predict(n)}predictProba(t){this.#i();let n=this.#t[this.#t.length-1].estimator;if(typeof n.predictProba!="function")throw new lt("Last step does not support predictProba");let e=this.#r(t);return n.predictProba(e)}score(t,n){this.#i();let e=this.#r(t);return this.#t[this.#t.length-1].estimator.score(e,n)}save(){this.#i();let t={typeId:Be,steps:this.#t.map(e=>({name:e.name,params:e.estimator.getParams()}))},n=this.#t.map(e=>({id:e.name,data:e.estimator.save(),mediaType:"application/x-wlearn-bundle"}));return wr(t,n)}static async load(t){let{manifest:n,toc:e,blobs:i}=yr(t),r=[];for(let a of n.steps){let l=e.find(f=>f.id===a.name);if(!l)throw new lt(`No artifact found for pipeline step "${a.name}"`);let h=i.subarray(l.offset,l.offset+l.length),c=await Le(h);r.push([a.name,c])}let o=new s(r);return o.#e=!0,o}dispose(){if(!this.#s){this.#s=!0;for(let t of this.#t)t.estimator.dispose()}}getParams(){let t={};for(let n of this.#t)t[n.name]=n.estimator.getParams();return t}setParams(t){this.#n();for(let n of this.#t)t[n.name]&&n.estimator.setParams(t[n.name]);return this}get capabilities(){return this.#t[this.#t.length-1].estimator.capabilities}get isFitted(){return this.#e}static registerLoader(){Oe||(Oe=!0,xr(Be,(t,n,e)=>s._loadFromParts(t,n,e)))}static async _loadFromParts(t,n,e){let i=[];for(let o of t.steps){let a=n.find(c=>c.id===o.name);if(!a)throw new lt(`No artifact found for pipeline step "${o.name}"`);let l=e.subarray(a.offset,a.offset+a.length),h=await Le(l);i.push([o.name,h])}let r=new s(i);return r.#e=!0,r}};ft.registerLoader();Ce.exports={Pipeline:ft}});var Xe=S((pc,De)=>{var{normalizeX:Te,normalizeY:mc}=G(),{ValidationError:$e}=z(),Rt=class s{#t;#e=!1;#s=null;#n=null;#i=null;#r=null;#o=0;constructor(t={}){this.#t={impute:t.impute??"auto",encode:t.encode??"auto",scale:t.scale??!1,maxCategories:t.maxCategories??20,...t}}fit(t,n){let e=Te(t),{data:i,rows:r,cols:o}=e;this.#s=new Array(o),this.#n=new Float64Array(o);let a=new Array(o);for(let h=0;h<o;h++){let c=[],f=!1,u=0,d=0;for(let w=0;w<r;w++){let b=i[w*o+h];Number.isNaN(b)?f=!0:(c.push(b),u+=b,d++)}let m=new Set(c),p=c.every(w=>w===Math.floor(w)),g=this.#t.encode!==!1&&p&&m.size<=this.#t.maxCategories&&m.size>=2;if(this.#s[h]=g?"categorical":"numeric",this.#t.impute!==!1&&f)if(g){let w=new Map;for(let _ of c)w.set(_,(w.get(_)||0)+1);let b=c[0],y=0;for(let[_,A]of w)A>y&&(b=_,y=A);this.#n[h]=b}else this.#n[h]=d>0?u/d:0;a[h]={values:c,unique:m,sum:u,count:d}}this.#i=new Array(o).fill(null);let l=0;if(this.#t.encode!==!1)for(let h=0;h<o;h++)if(this.#s[h]==="categorical"){let c=[...a[h].unique].sort((f,u)=>f-u);if(this.#t.encode==="label"||this.#t.encode===!1){let f=new Map;c.forEach((u,d)=>f.set(u,d)),this.#i[h]={type:"label",mapping:f,startIdx:l},l++}else{let f=new Map;c.forEach((u,d)=>f.set(u,l+d)),this.#i[h]={type:"onehot",mapping:f,startIdx:l,size:c.length},l+=c.length}}else l++;else l=o;if(this.#o=l,this.#r=new Array(this.#o).fill(null),this.#t.scale){let h=0;for(let c=0;c<o;c++){if(this.#s[c]==="categorical"&&this.#t.encode!==!1){let u=this.#i[c];h+=u.type==="onehot"?u.size:1;continue}let f=a[c].values;if(this.#t.scale==="standard"){let u=a[c].sum/(a[c].count||1),d=0;for(let p of f)d+=(p-u)**2;d/=f.length||1;let m=Math.sqrt(d);this.#r[h]={mean:u,std:m||1}}else if(this.#t.scale==="minmax"){let u=1/0,d=-1/0;for(let p of f)p<u&&(u=p),p>d&&(d=p);let m=d-u||1;this.#r[h]={min:u,range:m}}h++}}return this.#e=!0,this}transform(t){if(!this.#e)throw new $e("Preprocessor not fitted");let n=Te(t),{data:e,rows:i,cols:r}=n,o=this.#o,a=new Float64Array(i*o);for(let l=0;l<i;l++){let h=0;for(let c=0;c<r;c++){let f=e[l*r+c];if(Number.isNaN(f)&&this.#t.impute!==!1&&(f=this.#n[c]),this.#s[c]==="categorical"&&this.#i[c]){let u=this.#i[c];if(u.type==="onehot"){let d=u.mapping.get(f);d!==void 0&&(a[l*o+d]=1),h=u.startIdx+u.size}else{let d=u.mapping.get(f);a[l*o+h]=d!==void 0?d:-1,h++}}else a[l*o+h]=f,h++}if(this.#t.scale)for(let c=0;c<o;c++){let f=this.#r[c];f&&(this.#t.scale==="standard"?a[l*o+c]=(a[l*o+c]-f.mean)/f.std:this.#t.scale==="minmax"&&(a[l*o+c]=(a[l*o+c]-f.min)/f.range))}}return{data:a,rows:i,cols:o}}fitTransform(t,n){return this.fit(t,n),this.transform(t)}getState(){if(!this.#e)throw new $e("Preprocessor not fitted");return{config:this.#t,colTypes:this.#s,imputeValues:[...this.#n],encodings:this.#i.map(t=>t?{type:t.type,mapping:[...t.mapping.entries()],startIdx:t.startIdx,size:t.size}:null),scaleParams:this.#r,outputCols:this.#o}}static fromState(t){let n=new s(t.config);return n.#s=t.colTypes,n.#n=new Float64Array(t.imputeValues),n.#i=t.encodings.map(e=>e?{type:e.type,mapping:new Map(e.mapping),startIdx:e.startIdx,size:e.size}:null),n.#r=t.scaleParams,n.#o=t.outputCols,n.#e=!0,n}get isFitted(){return this.#e}get outputCols(){return this.#o}get capabilities(){return{transformer:!0}}dispose(){}};De.exports={Preprocessor:Rt}});var Qe=S((gc,Ze)=>{var{NotFittedError:Ve,DisposedError:Ue,ValidationError:J}=z(),{normalizeX:ht}=G(),{encodeBundle:Ye,encodeJSON:Ge,decodeJSON:We}=tt(),{register:Je}=ct(),Ke="wlearn.preprocess.standard_scaler@1",He="wlearn.preprocess.minmax_scaler@1",ut=class s{#t=null;#e=null;#s=!1;#n=!1;#i={};constructor(t={}){this.#i={...t}}fit(t){this.#r();let{rows:n,cols:e,data:i}=ht(t);if(n===0)throw new J("Cannot fit on empty data");let r=new Float64Array(e),o=new Float64Array(e);for(let l=0;l<n;l++){let h=l+1;for(let c=0;c<e;c++){let f=i[l*e+c],u=f-r[c];r[c]+=u/h;let d=f-r[c];o[c]+=u*d}}let a=new Float64Array(e);for(let l=0;l<e;l++)a[l]=n>1?Math.sqrt(o[l]/(n-1)):0;return this.#t=r,this.#e=a,this.#s=!0,this}transform(t){this.#o();let{rows:n,cols:e,data:i}=ht(t);if(e!==this.#t.length)throw new J(`Expected ${this.#t.length} columns, got ${e}`);let r=new Float64Array(n*e);for(let o=0;o<n;o++)for(let a=0;a<e;a++){let l=o*e+a,h=this.#e[a];r[l]=h>0?(i[l]-this.#t[a])/h:0}return{rows:n,cols:e,data:r}}fitTransform(t){return this.fit(t),this.transform(t)}save(){this.#o();let t={means:Array.from(this.#t),stds:Array.from(this.#e)};return Ye({typeId:Ke,params:this.getParams()},[{id:"params",data:Ge(t),mediaType:"application/json"}])}static _fromBundle(t,n,e){let i=n.find(a=>a.id==="params");if(!i)throw new J('Bundle missing "params" artifact');let r=We(e.subarray(i.offset,i.offset+i.length)),o=new s(t.params||{});return o.#t=new Float64Array(r.means),o.#e=new Float64Array(r.stds),o.#s=!0,o}dispose(){this.#n||(this.#n=!0,this.#t=null,this.#e=null,this.#s=!1)}getParams(){return{...this.#i}}setParams(t){return Object.assign(this.#i,t),this}get isFitted(){return this.#s&&!this.#n}#r(){if(this.#n)throw new Ue("StandardScaler has been disposed.")}#o(){if(this.#r(),!this.#s)throw new Ve("StandardScaler is not fitted.")}},dt=class s{#t=null;#e=null;#s=!1;#n=!1;#i={};constructor(t={}){this.#i={...t}}fit(t){this.#r();let{rows:n,cols:e,data:i}=ht(t);if(n===0)throw new J("Cannot fit on empty data");let r=new Float64Array(e).fill(1/0),o=new Float64Array(e).fill(-1/0);for(let a=0;a<n;a++)for(let l=0;l<e;l++){let h=i[a*e+l];h<r[l]&&(r[l]=h),h>o[l]&&(o[l]=h)}return this.#t=r,this.#e=o,this.#s=!0,this}transform(t){this.#o();let{rows:n,cols:e,data:i}=ht(t);if(e!==this.#t.length)throw new J(`Expected ${this.#t.length} columns, got ${e}`);let r=new Float64Array(n*e);for(let o=0;o<n;o++)for(let a=0;a<e;a++){let l=o*e+a,h=this.#e[a]-this.#t[a];r[l]=h>0?(i[l]-this.#t[a])/h:0}return{rows:n,cols:e,data:r}}fitTransform(t){return this.fit(t),this.transform(t)}save(){this.#o();let t={mins:Array.from(this.#t),maxs:Array.from(this.#e)};return Ye({typeId:He,params:this.getParams()},[{id:"params",data:Ge(t),mediaType:"application/json"}])}static _fromBundle(t,n,e){let i=n.find(a=>a.id==="params");if(!i)throw new J('Bundle missing "params" artifact');let r=We(e.subarray(i.offset,i.offset+i.length)),o=new s(t.params||{});return o.#t=new Float64Array(r.mins),o.#e=new Float64Array(r.maxs),o.#s=!0,o}dispose(){this.#n||(this.#n=!0,this.#t=null,this.#e=null,this.#s=!1)}getParams(){return{...this.#i}}setParams(t){return Object.assign(this.#i,t),this}get isFitted(){return this.#s&&!this.#n}#r(){if(this.#n)throw new Ue("MinMaxScaler has been disposed.")}#o(){if(this.#r(),!this.#s)throw new Ve("MinMaxScaler is not fitted.")}};Je(Ke,ut._fromBundle);Je(He,dt._fromBundle);Ze.exports={StandardScaler:ut,MinMaxScaler:dt}});var Mt=S((bc,ts)=>{function Sr(s=42){let t=s|0;return()=>(t=t*1664525+1013904223&2147483647,t/2147483647)}function _r(s,t){for(let n=s.length-1;n>0;n--){let e=t()*(n+1)|0,i=s[n];s[n]=s[e],s[e]=i}return s}ts.exports={makeLCG:Sr,shuffle:_r}});var Lt=S((wc,es)=>{var{ValidationError:D}=z();function X(s,t,n){if(!s||!t||s.length===0||t.length===0)throw new D(`${n}: inputs must be non-empty`);if(s.length!==t.length)throw new D(`${n}: length mismatch (${s.length} vs ${t.length})`)}function K(s,t){let n=new Set;for(let r=0;r<s.length;r++)n.add(s[r]);if(t)for(let r=0;r<t.length;r++)n.add(t[r]);let e=new Int32Array([...n].sort((r,o)=>r-o)),i=new Map;for(let r=0;r<e.length;r++)i.set(e[r],r);return{labels:e,labelMap:i,nClasses:e.length}}function mt(s,t,n,e){let i=new Int32Array(e*e);for(let r=0;r<s.length;r++){let o=n.get(s[r]),a=n.get(t[r]);i[o*e+a]++}return i}function zt(s,t){let n=new Float64Array(t),e=new Float64Array(t),i=new Float64Array(t);for(let r=0;r<t;r++){n[r]=s[r*t+r];for(let o=0;o<t;o++)o!==r&&(e[r]+=s[o*t+r],i[r]+=s[r*t+o])}return{tp:n,fp:e,fn:i}}function vr(s,t){X(s,t,"accuracy");let n=0;for(let e=0;e<s.length;e++)s[e]===t[e]&&n++;return n/s.length}function Fr(s,t){X(s,t,"r2Score");let n=s.length,e=0;for(let o=0;o<n;o++)e+=s[o];e/=n;let i=0,r=0;for(let o=0;o<n;o++){let a=s[o]-e;i+=a*a;let l=s[o]-t[o];r+=l*l}return i===0?0:1-r/i}function Ar(s,t){X(s,t,"meanSquaredError");let n=0;for(let e=0;e<s.length;e++){let i=s[e]-t[e];n+=i*i}return n/s.length}function kr(s,t){X(s,t,"meanAbsoluteError");let n=0;for(let e=0;e<s.length;e++)n+=Math.abs(s[e]-t[e]);return n/s.length}function qr(s,t){X(s,t,"confusionMatrix");let{labels:n,labelMap:e,nClasses:i}=K(s,t);return{matrix:mt(s,t,e,i),labels:n}}function Nt(s,t,n){if(s==="binary"){if(t>2)throw new D('average="binary" requires exactly 2 classes');return"binary"}return s||"binary"}function Er(s,t,{average:n="binary"}={}){X(s,t,"precisionScore");let{labels:e,labelMap:i,nClasses:r}=K(s,t),o=mt(s,t,i,r),{tp:a,fp:l}=zt(o,r),h=Nt(n,r,e);if(h==="binary"){let f=r-1,u=a[f]+l[f];return u===0?0:a[f]/u}if(h==="micro"){let f=0,u=0;for(let d=0;d<r;d++)f+=a[d],u+=l[d];return f+u===0?0:f/(f+u)}let c=0;for(let f=0;f<r;f++){let u=a[f]+l[f];c+=u===0?0:a[f]/u}return c/r}function Ir(s,t,{average:n="binary"}={}){X(s,t,"recallScore");let{labels:e,labelMap:i,nClasses:r}=K(s,t),o=mt(s,t,i,r),{tp:a,fn:l}=zt(o,r),h=Nt(n,r,e);if(h==="binary"){let f=r-1,u=a[f]+l[f];return u===0?0:a[f]/u}if(h==="micro"){let f=0,u=0;for(let d=0;d<r;d++)f+=a[d],u+=l[d];return f+u===0?0:f/(f+u)}let c=0;for(let f=0;f<r;f++){let u=a[f]+l[f];c+=u===0?0:a[f]/u}return c/r}function Pr(s,t,{average:n="binary"}={}){X(s,t,"f1Score");let{labels:e,labelMap:i,nClasses:r}=K(s,t),o=mt(s,t,i,r),{tp:a,fp:l,fn:h}=zt(o,r),c=Nt(n,r,e);function f(d,m,p){let g=d+m===0?0:d/(d+m),w=d+p===0?0:d/(d+p);return g+w===0?0:2*g*w/(g+w)}if(c==="binary"){let d=r-1;return f(a[d],l[d],h[d])}if(c==="micro"){let d=0,m=0,p=0;for(let g=0;g<r;g++)d+=a[g],m+=l[g],p+=h[g];return f(d,m,p)}let u=0;for(let d=0;d<r;d++)u+=f(a[d],l[d],h[d]);return u/r}function Rr(s,t,{nClasses:n,eps:e=1e-15}={}){if(!s||s.length===0)throw new D("logLoss: yTrue must be non-empty");let i=s.length;if(!n){let a=new Set;for(let l=0;l<i;l++)a.add(s[l]);n=a.size}if(t.length!==i*n)throw new D(`logLoss: yProba length (${t.length}) must be n * nClasses (${i*n})`);let{labelMap:r}=K(s),o=0;for(let a=0;a<i;a++){let l=r.get(s[a]),h=t[a*n+l];h=Math.max(e,Math.min(1-e,h)),o-=Math.log(h)}return o/i}function Mr(s,t){if(!s||s.length===0)throw new D("rocAuc: yTrue must be non-empty");if(s.length!==t.length)throw new D("rocAuc: length mismatch");let{labels:n}=K(s);if(n.length!==2)throw new D("rocAuc: requires exactly 2 classes");let e=n[n.length-1],i=s.length,r=Array.from({length:i},(d,m)=>m);r.sort((d,m)=>t[m]-t[d]);let o=0,a=0;for(let d=0;d<i;d++)s[d]===e?o++:a++;if(o===0||a===0)return 0;let l=0,h=0,c=0,f=0,u=0;for(let d=0;d<i;d++){let m=r[d];s[m]===e?l++:h++;let p=l/o,g=h/a;u+=(g-f)*(p+c)/2,c=p,f=g}return u}es.exports={accuracy:vr,r2Score:Fr,meanSquaredError:Ar,meanAbsoluteError:kr,confusionMatrix:qr,precisionScore:Er,recallScore:Ir,f1Score:Pr,logLoss:Rr,rocAuc:Mr}});var cs=S((yc,as)=>{var{ValidationError:U}=z(),{makeLCG:Bt,shuffle:Ot}=Mt(),{normalizeX:zr,normalizeY:Nr}=G(),{accuracy:Lr,r2Score:Br,meanSquaredError:Or,meanAbsoluteError:Cr}=Lt(),ss={accuracy:(s,t)=>Lr(s,t),r2:(s,t)=>Br(s,t),neg_mse:(s,t)=>-Or(s,t),neg_mae:(s,t)=>-Cr(s,t)};function is(s){if(typeof s=="function")return s;let t=ss[s];if(!t)throw new U(`Unknown scoring: "${s}". Available: ${Object.keys(ss).join(", ")}`);return t}function jr(s,t=5,{shuffle:n=!0,seed:e=42}={}){if(s<t)throw new U(`kFold: n (${s}) must be >= k (${t})`);if(t<2)throw new U("kFold: k must be >= 2");let i=Int32Array.from({length:s},(h,c)=>c);if(n){let h=Bt(e);Ot(i,h)}let r=Math.floor(s/t),o=s%t,a=[],l=0;for(let h=0;h<t;h++){let c=r+(h<o?1:0),f=i.slice(l,l+c),u=[];l>0&&u.push(i.slice(0,l)),l+c<s&&u.push(i.slice(l+c));let d=Dr(u);a.push({train:d,test:f}),l+=c}return a}function os(s,t=5,{shuffle:n=!0,seed:e=42}={}){let i=s.length;if(i<t)throw new U(`stratifiedKFold: n (${i}) must be >= k (${t})`);if(t<2)throw new U("stratifiedKFold: k must be >= 2");let r=new Map;for(let h=0;h<i;h++){let c=s[h];r.has(c)||r.set(c,[]),r.get(c).push(h)}if(n){let h=Bt(e);for(let c of r.values())Ot(c,h)}let o=Array.from({length:t},()=>[]);for(let h of r.values())for(let c=0;c<h.length;c++)o[c%t].push(h[c]);let a=Int32Array.from({length:i},(h,c)=>c),l=[];for(let h=0;h<t;h++){let c=new Set(o[h]),f=new Int32Array(o[h]),u=a.filter(d=>!c.has(d));l.push({train:u,test:f})}return l}function Tr(s,{testSize:t=.2,shuffle:n=!0,seed:e=42}={}){if(s<2)throw new U("trainTestSplit: n must be >= 2");let i=Math.max(1,Math.round(s*t)),r=s-i;if(r<1)throw new U("trainTestSplit: testSize too large");let o=Int32Array.from({length:s},(a,l)=>l);if(n){let a=Bt(e);Ot(o,a)}return{train:o.slice(0,r),test:o.slice(r)}}async function $r(s,t,n,{cv:e=5,scoring:i="accuracy",seed:r=42,params:o={}}={}){let a=zr(t),l=Nr(n),h=is(i),c;Array.isArray(e)?c=e:c=os(l,e,{shuffle:!0,seed:r});let f=new Float64Array(c.length);for(let u=0;u<c.length;u++){let{train:d,test:m}=c[u],p=ns(a,d),g=rs(l,d),w=ns(a,m),b=rs(l,m),y=await s.create(o);try{y.fit(p,g);let _=await y.predict(w);f[u]=h(b,_)}finally{y.dispose()}}return f}function Dr(s){let t=0;for(let i of s)t+=i.length;let n=new Int32Array(t),e=0;for(let i of s)n.set(i,e),e+=i.length;return n}function ns(s,t){let{data:n,cols:e}=s,i=t.length,r=new Float64Array(i*e);for(let o=0;o<i;o++){let a=t[o]*e;r.set(n.subarray(a,a+e),o*e)}return{data:r,rows:i,cols:e}}function rs(s,t){let n=new s.constructor(t.length);for(let e=0;e<t.length;e++)n[e]=s[t[e]];return n}as.exports={kFold:jr,stratifiedKFold:os,trainTestSplit:Tr,crossValScore:$r,getScorer:is}});var hs=S((xc,fs)=>{function ls(s){return s!=null&&typeof s.then=="function"}function Xr(s,t){return ls(s)?s.then(t):t(s)}fs.exports={isPromiseLike:ls,lift:Xr}});var ps=S((Sc,ms)=>{var{normalizeY:Vr}=G(),{round:Ur}=Math;function us(s){let t=Vr(s);if(t instanceof Int32Array)return"classification";let n=new Set;for(let e=0;e<t.length;e++){if(t[e]!==Ur(t[e]))return"regression";n.add(t[e])}return n.size<=20?"classification":"regression"}var ds=new WeakMap;function I(s){let t=ds.get(s);if(!t)throw new Error("Model: invalid instance");return t}function et(s,t){let n=I(s);if(n.disposed)throw new Error(`${t} has been disposed.`);if(!n.inner)throw new Error(`${t}: not fitted`);return n.inner}function Yr(s,t,n={}){let e=n.name||"Model",i=n.load||null,r=s===t;class o{constructor(c,f){ds.set(this,{inner:null,task:c,params:f,disposed:!1})}static async create(c={}){let f={...c},u=f.task||null;delete f.task,i&&await i();let d=new o(u,f),m=I(d);if(u){let p=u==="classification"?s:t;m.inner=await p.create({...f,task:u})}else r&&(m.inner=await s.create(f));return d}static async load(c){try{let f=await s.load(c),u=new o("classification",{}),d=I(u);return d.inner=f,typeof f.getParams=="function"&&(d.params=f.getParams(),d.params.task&&(d.task=d.params.task)),u}catch{let u=await t.load(c),d=new o("regression",{}),m=I(d);return m.inner=u,typeof u.getParams=="function"&&(m.params=u.getParams(),m.params.task&&(m.task=m.params.task)),d}}fit(c,f,u){let d=I(this);if(d.task||(d.task=us(f)),!d.inner){let m=d.task==="classification"?s:t;d.inner=new m({...d.params,task:d.task})}return d.inner.fit(c,f,u),this}predict(c,f){return et(this,e).predict(c,f)}predictProba(c,f){let u=et(this,e);if(typeof u.predictProba!="function")throw new Error(`${e}: predictProba not available`);return u.predictProba(c,f)}score(c,f,u){return et(this,e).score(c,f,u)}save(){return et(this,e).save()}dispose(){let c=I(this);c.inner&&typeof c.inner.dispose=="function"&&c.inner.dispose(),c.inner=null,c.disposed=!0}getParams(){let c=I(this),f=c.inner&&typeof c.inner.getParams=="function"?c.inner.getParams():{...c.params};return c.task&&(f.task=c.task),f}setParams(c){let f=I(this),u={...c};if(u.task){let d=u.task;delete u.task,d!==f.task&&f.inner&&this.dispose(),f.task=d}return f.params=u,f.inner&&typeof f.inner.setParams=="function"&&f.inner.setParams(u),this}get task(){return I(this).task}get isFitted(){let c=I(this);return c.inner?c.inner.isFitted!==void 0?c.inner.isFitted:!0:!1}get classes(){let c=I(this);return c.inner?typeof c.inner.classes=="function"?c.inner.classes():c.inner.classes!==void 0?c.inner.classes:new Int32Array(0):new Int32Array(0)}get capabilities(){let c=I(this);return c.inner&&c.inner.capabilities!==void 0?c.inner.capabilities:{}}}let a=new Set(["constructor","fit","predict","predictProba","score","save","dispose","getParams","setParams"]),l=new Set(["classes","task","isFitted","capabilities"]);for(let h of[s,t])if(!(!h||!h.prototype))for(let c of Object.getOwnPropertyNames(h.prototype)){if(c.startsWith("_")||c.startsWith("#")||a.has(c)||l.has(c)||c in o.prototype)continue;let f=Object.getOwnPropertyDescriptor(h.prototype,c);f&&(typeof f.value=="function"?o.prototype[c]=function(...u){return et(this,e)[c](...u)}:f.get&&Object.defineProperty(o.prototype,c,{get(){let u=I(this);if(u.inner)return u.inner[c]},configurable:!0}))}return(s.defaultSearchSpace||t.defaultSearchSpace)&&(o.defaultSearchSpace=()=>s.defaultSearchSpace?.()||t.defaultSearchSpace?.()||{}),(s.budgetSpec||t.budgetSpec)&&(o.budgetSpec=()=>s.budgetSpec?.()||t.budgetSpec?.()||void 0),Object.defineProperty(o,"name",{value:e,configurable:!0}),o}ms.exports={createModelClass:Yr,detectTask:us}});var q=S((_c,gs)=>{var{WlearnError:Gr,BundleError:Wr,RegistryError:Jr,ValidationError:Kr,NotFittedError:Hr,DisposedError:Zr}=z(),{normalizeX:Qr,normalizeY:ti,makeDense:ei,validateMatrix:si}=G(),{sha256Sync:ni}=kt(),{encodeBundle:ri,decodeBundle:ii,validateBundle:oi,encodeJSON:ai,decodeJSON:ci}=tt(),{register:li,load:fi,loadSync:hi,getRegistry:ui}=ct(),{Pipeline:di}=je(),{Step:mi}=Pt(),{Preprocessor:pi}=Xe(),{StandardScaler:gi,MinMaxScaler:bi}=Qe(),{makeLCG:wi,shuffle:yi}=Mt(),{accuracy:xi,r2Score:Si,meanSquaredError:_i,meanAbsoluteError:vi,confusionMatrix:Fi,precisionScore:Ai,recallScore:ki,f1Score:qi,logLoss:Ei,rocAuc:Ii}=Lt(),{kFold:Pi,stratifiedKFold:Ri,trainTestSplit:Mi,crossValScore:zi,getScorer:Ni}=cs(),{isPromiseLike:Li,lift:Bi}=hs(),{createModelClass:Oi,detectTask:Ci}=ps();gs.exports={WlearnError:Gr,BundleError:Wr,RegistryError:Jr,ValidationError:Kr,NotFittedError:Hr,DisposedError:Zr,normalizeX:Qr,normalizeY:ti,makeDense:ei,validateMatrix:si,sha256Sync:ni,encodeBundle:ri,decodeBundle:ii,validateBundle:oi,encodeJSON:ai,decodeJSON:ci,register:li,load:fi,loadSync:hi,getRegistry:ui,Pipeline:di,Step:mi,Preprocessor:pi,StandardScaler:gi,MinMaxScaler:bi,makeLCG:wi,shuffle:yi,accuracy:xi,r2Score:Si,meanSquaredError:_i,meanAbsoluteError:vi,confusionMatrix:Fi,precisionScore:Ai,recallScore:ki,f1Score:qi,logLoss:Ei,rocAuc:Ii,kFold:Pi,stratifiedKFold:Ri,trainTestSplit:Mi,crossValScore:zi,getScorer:Ni,isPromiseLike:Li,lift:Bi,createModelClass:Oi,detectTask:Ci}});var st=S((Fc,ys)=>{var{makeLCG:ji}=q(),{floor:Ct,round:jt,log:B,exp:gt,min:vc,max:pt}=Math;function Tt(s,t){let{type:n}=s;switch(n){case"categorical":return s.values[Ct(t()*s.values.length)];case"uniform":return s.low+t()*(s.high-s.low);case"log_uniform":return gt(B(s.low)+t()*(B(s.high)-B(s.low)));case"int_uniform":return s.low+Ct(t()*(s.high-s.low+1));case"int_log_uniform":return jt(gt(B(s.low)+t()*(B(s.high)-B(s.low))));default:throw new Error(`Unknown SearchParam type: "${n}"`)}}function ws(s,t){let n={},e=Object.keys(s);for(let i of e)s[i].condition||(n[i]=Tt(s[i],t));for(let i of e){let{condition:r}=s[i];if(!r)continue;let o=!0;for(let[a,l]of Object.entries(r))if(n[a]!==l){o=!1;break}o&&(n[i]=Tt(s[i],t))}return n}function Ti(s,t,{seed:n=42}={}){let e=ji(n),i=[];for(let r=0;r<t;r++)i.push(ws(s,e));return i}function $i(s,{steps:t=5}={}){let n=Object.keys(s);if(n.length===0)return[{}];let e=n.filter(a=>!s[a].condition),i=n.filter(a=>s[a].condition),r=e.map(a=>bs(s[a],t)),o=[{}];for(let a=0;a<e.length;a++){let l=e[a],h=r[a],c=[];for(let f of o)for(let u of h)c.push({...f,[l]:u});o=c}for(let a of o)for(let l of i){let{condition:h}=s[l],c=!0;for(let[f,u]of Object.entries(h))if(a[f]!==u){c=!1;break}if(c){let f=bs(s[l],t);a[l]=f[Ct(f.length/2)]}}return o}function bs(s,t){let{type:n}=s;switch(n){case"categorical":return[...s.values];case"uniform":{let e=[];for(let i=0;i<t;i++)e.push(s.low+(s.high-s.low)*i/pt(1,t-1));return e}case"log_uniform":{let e=B(s.low),i=B(s.high),r=[];for(let o=0;o<t;o++)r.push(gt(e+(i-e)*o/pt(1,t-1)));return r}case"int_uniform":{if(s.high-s.low+1<=t){let r=[];for(let o=s.low;o<=s.high;o++)r.push(o);return r}let i=[];for(let r=0;r<t;r++)i.push(s.low+jt((s.high-s.low)*r/pt(1,t-1)));return[...new Set(i)].sort((r,o)=>r-o)}case"int_log_uniform":{let e=B(s.low),i=B(s.high),r=[];for(let o=0;o<t;o++)r.push(jt(gt(e+(i-e)*o/pt(1,t-1))));return[...new Set(r)].sort((o,a)=>o-a)}default:throw new Error(`Unknown SearchParam type: "${n}"`)}}ys.exports={sampleParam:Tt,sampleConfig:ws,randomConfigs:Ti,gridConfigs:$i}});var Dt=S((Ac,xs)=>{var $t=class s{#t=[];#e=0;#s=!0;add({modelName:t,params:n,scores:e,fitTimeMs:i}){let r=0;for(let c=0;c<e.length;c++)r+=e[c];let o=r/e.length,a=0;for(let c=0;c<e.length;c++){let f=e[c]-o;a+=f*f}let l=Math.sqrt(a/e.length),h={id:this.#e++,modelName:t,params:n,scores:e,meanScore:o,stdScore:l,fitTimeMs:i,rank:0};return this.#t.push(h),this.#s=!0,h}ranked(){if(this.#s){this.#t.sort((t,n)=>n.meanScore-t.meanScore);for(let t=0;t<this.#t.length;t++)this.#t[t].rank=t+1;this.#s=!1}return this.#t.slice()}best(){return this.#t.length===0?null:(this.ranked(),this.#t[0])}top(t){return this.ranked().slice(0,t)}toJSON(){return this.ranked().map(t=>({id:t.id,modelName:t.modelName,params:t.params,scores:[...t.scores],meanScore:t.meanScore,stdScore:t.stdScore,fitTimeMs:t.fitTimeMs,rank:t.rank}))}static fromJSON(t){let n=new s;for(let e of t)n.#t.push({...e,scores:new Float64Array(e.scores)}),e.id>=n.#e&&(n.#e=e.id+1);return n.#s=!0,n}get length(){return this.#t.length}};xs.exports={Leaderboard:$t}});var M=S((qc,Ss)=>{var{makeLCG:kc}=q(),{round:Di}=Math;function Xi(s){if(s instanceof Int32Array)return"classification";let t=new Set;for(let n=0;n<s.length;n++){if(s[n]!==Di(s[n]))return"regression";t.add(s[n])}return t.size<=20?"classification":"regression"}function Vi(){return typeof performance<"u"?performance.now():Date.now()}function Xt(s){return s==null?String(s):typeof s=="number"?s.toString():typeof s=="string"?JSON.stringify(s):typeof s=="boolean"?String(s):Array.isArray(s)?"["+s.map(Xt).join(",")+"]":typeof s=="object"?"{"+Object.keys(s).sort().map(n=>JSON.stringify(n)+":"+Xt(s[n])).join(",")+"}":String(s)}function Ui(s,t){return s+":"+Xt(t)}function Yi(s){let t=2166136261;for(let n=0;n<s.length;n++)t^=s.charCodeAt(n),t=t*16777619&2147483647;return t}function Gi(s,t,n){let e=Yi(s),i=n*2654435761+e*40503+t*65537&2147483647;return i=(i>>>16^i)*73244475&2147483647,i}function Wi(s,t,n){let e=s.length,i=Math.min(t,e);for(let r=0;r<i;r++){let o=r+(n()*(e-r)|0),a=s[r];s[r]=s[o],s[o]=a}return s.subarray?s.subarray(0,i):s.slice(0,i)}function Ji(s){if(typeof s=="function")return!0;switch(s){case"accuracy":case"r2":case"neg_mse":case"neg_mae":return!0;default:return!0}}Ss.exports={detectTask:Xi,now:Vi,makeCandidateId:Ui,seedFor:Gi,partialShuffle:Wi,scorerGreaterIsBetter:Ji}});var H=S((Rc,Fs)=>{var{normalizeX:Ec,normalizeY:Ic,makeLCG:Ki,getScorer:Hi}=q(),{Leaderboard:Zi}=Dt(),{now:nt,seedFor:Qi,partialShuffle:to}=M(),{ceil:eo,min:Pc}=Math;function _s(s,t){let{data:n,cols:e}=s,i=t.length,r=new Float64Array(i*e);for(let o=0;o<i;o++){let a=t[o]*e;r.set(n.subarray(a,a+e),o*e)}return{data:r,rows:i,cols:e}}function vs(s,t){let n=new s.constructor(t.length);for(let e=0;e<t.length;e++)n[e]=s[t[e]];return n}var Vt=class{#t;#e;#s;#n;#i;#r;#o;#l;#a;constructor({folds:t,scoring:n,X:e,y:i,timeLimitMs:r=0,seed:o=42,onProgress:a}){this.#t=t,this.#e=Hi(n),this.#s=e,this.#n=i,this.#i=r,this.#r=o,this.#o=nt(),this.#l=new Zi,this.#a=a||null}get leaderboard(){return this.#l}get isTimedOut(){return this.#i<=0?!1:nt()-this.#o>this.#i}async evaluateCandidate({candidateId:t,cls:n,params:e,budget:i}){let r=this.#t,o=new Float64Array(r.length),a=nt(),l=0,h=this.#c(n,e,i);for(let u=0;u<r.length;u++){let{train:d,test:m}=r[u];i&&i.type==="subsample"&&(d=this.#f(d,i.value,t,u)),l+=d.length;let p=_s(this.#s,d),g=vs(this.#n,d),w=_s(this.#s,m),b=vs(this.#n,m),y=await n.create(h);try{y.fit(p,g);let _=await y.predict(w);o[u]=this.#e(b,_)}finally{y.dispose()}}let c=nt()-a,f=this.#l.add({modelName:t.split(":")[0],params:e,scores:o,fitTimeMs:c});return{candidateId:t,meanScore:f.meanScore,foldScores:o,stdScore:f.stdScore,fitTimeMs:c,nTrainUsed:Math.round(l/r.length),nTest:r[0].test.length}}#c(t,n,e){if(!e||e.type!=="rounds")return n;let i=t.budgetSpec?.();return!i||!i.roundsParam||n[i.roundsParam]!==void 0?n:{...n,[i.roundsParam]:e.value}}#f(t,n,e,i){let r=Math.max(1,eo(t.length*n));if(r>=t.length)return t;let o=new Int32Array(t),a=Qi(e,i,this.#r),l=Ki(a);return to(o,r,l)}async runStrategy(t){let n=0;for(;!t.isDone()&&!this.isTimedOut;){let e=t.next();if(e===null)break;try{let i=await this.evaluateCandidate(e);if(t.report(i),n++,this.#a){let r=this.#l.best();this.#a({phase:"search",candidatesDone:n,bestScore:r?r.meanScore:null,bestModel:r?r.modelName:null,lastCandidate:{model:i.candidateId.split(":")[0],score:i.meanScore,timeMs:i.fitTimeMs},elapsedMs:nt()-this.#o})}}catch{n++}}return{leaderboard:this.#l}}};Fs.exports={Executor:Vt}});var Yt=S((Mc,ks)=>{var{makeLCG:As}=q(),{sampleConfig:so}=st(),{makeCandidateId:no}=M(),Ut=class{#t=[];#e=0;#s=0;constructor(t,{nIter:n=20,seed:e=42}={}){let i=As(e);for(let r of t){let a={...r.searchSpace||r.cls.defaultSearchSpace?.()||{}};if(r.params)for(let h of Object.keys(r.params))delete a[h];let l=As(i()*2147483647|0);for(let h=0;h<n;h++){let f={...so(a,l),...r.params||{}},u=no(r.name,f);this.#t.push({candidateId:u,cls:r.cls,params:f})}}this.#s=this.#t.length}next(){return this.#e>=this.#s?null:this.#t[this.#e++]}report(t){}isDone(){return this.#e>=this.#s}};ks.exports={RandomStrategy:Ut}});var Jt=S((Nc,Is)=>{var{stratifiedKFold:ro,kFold:io,normalizeX:qs,normalizeY:Es,ValidationError:Gt}=q(),{Executor:oo}=H(),{RandomStrategy:ao}=Yt(),{detectTask:co,scorerGreaterIsBetter:zc}=M(),Wt=class{#t;#e;#s=null;#n=null;constructor(t,n={}){if(!t||t.length===0)throw new Gt("RandomSearch: at least one model is required");this.#t=t,this.#e={scoring:null,cv:5,seed:42,task:null,nIter:20,maxTimeMs:0,onProgress:null,...n}}async fit(t,n){let e=qs(t),i=Es(n),r=this.#e.task||co(i),o=this.#e.scoring||(r==="classification"?"accuracy":"r2"),{cv:a,seed:l,nIter:h,maxTimeMs:c,onProgress:f}=this.#e,u=r==="classification"?ro(i,a,{shuffle:!0,seed:l}):io(i.length,a,{shuffle:!0,seed:l}),d=new oo({folds:u,scoring:o,X:e,y:i,timeLimitMs:c,seed:l,onProgress:f}),m=new ao(this.#t,{nIter:h,seed:l}),{leaderboard:p}=await d.runStrategy(m);if(p.length===0)throw new Gt("RandomSearch: no candidates were evaluated");return this.#s=p,this.#n=p.best(),{leaderboard:p,bestResult:this.#n}}async refitBest(t,n){if(!this.#n)throw new Gt("RandomSearch: must call fit() first");let e=this.#n,r=await this.#t.find(l=>l.name===e.modelName).cls.create(e.params),o=qs(t),a=Es(n);return r.fit(o,a),r}get leaderboard(){return this.#s}get bestResult(){return this.#n}};Is.exports={RandomSearch:Wt}});var Qt=S((Lc,zs)=>{var{makeLCG:Ps}=q(),{sampleConfig:lo}=st(),{makeCandidateId:fo}=M(),{ceil:Rs,log:Ms,max:Kt,min:bt,floor:Ht}=Math,Zt=class{#t;#e=0;#s=0;#n;#i;#r;#o;#l;#a=[];#c=[];#f=!1;#h=!1;constructor(t,{nIter:n=20,seed:e=42,factor:i=3,nSamples:r,greaterIsBetter:o=!0,cv:a=5}={}){this.#i=i,this.#r=r||0,this.#l=o;let l=Ps(e),h=[];for(let c of t){let u={...c.searchSpace||c.cls.defaultSearchSpace?.()||{}};if(c.params)for(let m of Object.keys(c.params))delete u[m];let d=Ps(l()*2147483647|0);for(let m=0;m<n;m++){let g={...lo(u,d),...c.params||{}},w=fo(c.name,g);h.push({candidateId:w,cls:c.cls,params:g})}}this.#t=h,this.#n=Kt(1,Rs(Ms(h.length)/Ms(i))),this.#o=Kt(a*2,Ht(this.#r/i**this.#n))}next(){if(this.#f||this.#e>=this.#t.length)return null;let t=this.#t[this.#e++];if(!this.#h){let n=bt(this.#r,Ht(this.#o*this.#i**this.#s)),e=bt(1,n/this.#r);if(e<1)return{...t,budget:{type:"subsample",value:e}}}return t}report(t){if(this.#a.push(t),this.#a.length<this.#t.length)return;if(this.#h){this.#f=!0;return}let n=bt(this.#r,Ht(this.#o*this.#i**this.#s)),e=bt(1,n/this.#r),i=[...this.#a];this.#l?i.sort((a,l)=>l.meanScore-a.meanScore):i.sort((a,l)=>a.meanScore-l.meanScore);let r=Kt(1,Rs(i.length/this.#i));this.#c.push({round:this.#s,nResources:n,fraction:e,nCandidates:this.#a.length,nSurvivors:r});let o=new Set(i.slice(0,r).map(a=>a.candidateId));this.#t=this.#t.filter(a=>o.has(a.candidateId)),this.#s++,this.#e=0,this.#a=[],(this.#t.length<=1||this.#s>=this.#n)&&(this.#h=!0)}isDone(){return this.#f}get rounds(){return this.#c}};zs.exports={HalvingStrategy:Zt}});var ee=S((Bc,Os)=>{var{stratifiedKFold:ho,kFold:uo,normalizeX:Ns,normalizeY:Ls,ValidationError:Bs}=q(),{Executor:mo}=H(),{HalvingStrategy:po}=Qt(),{detectTask:go,scorerGreaterIsBetter:bo}=M(),te=class{#t;#e;#s=null;#n=null;#i=null;constructor(t,n={}){if(!t||t.length===0)throw new Bs("SuccessiveHalvingSearch: at least one model is required");this.#t=t,this.#e={scoring:null,cv:5,seed:42,task:null,nIter:20,maxTimeMs:0,factor:3,minResources:0,onProgress:null,...n}}async fit(t,n){let e=Ns(t),i=Ls(n),r=e.rows,o=this.#e.task||go(i),a=this.#e.scoring||(o==="classification"?"accuracy":"r2"),l=bo(a),{cv:h,seed:c,nIter:f,maxTimeMs:u,factor:d,onProgress:m}=this.#e,p=o==="classification"?ho(i,h,{shuffle:!0,seed:c}):uo(r,h,{shuffle:!0,seed:c}),g=new mo({folds:p,scoring:a,X:e,y:i,timeLimitMs:u,seed:c,onProgress:m}),w=new po(this.#t,{nIter:f,seed:c,factor:d,nSamples:r,greaterIsBetter:l,cv:h}),{leaderboard:b}=await g.runStrategy(w);return this.#s=b,this.#n=b.best(),this.#i=w.rounds,{leaderboard:b,bestResult:this.#n,rounds:this.#i}}async refitBest(t,n){if(!this.#n)throw new Bs("SuccessiveHalvingSearch: must call fit() first");let e=this.#n,r=await this.#t.find(l=>l.name===e.modelName).cls.create(e.params),o=Ns(t),a=Ls(n);return r.fit(o,a),r}get leaderboard(){return this.#s}get bestResult(){return this.#n}get rounds(){return this.#i}};Os.exports={SuccessiveHalvingSearch:te}});var ie=S((Oc,$s)=>{var{stratifiedKFold:wo,kFold:yo,normalizeX:Cs,normalizeY:js,ValidationError:se}=q(),{Executor:xo}=H(),{detectTask:So,makeCandidateId:_o}=M(),ne={classification:{xgb:[{objective:"multi:softprob",eta:.05,max_depth:6,numRound:200,subsample:.8,colsample_bytree:.8,min_child_weight:1,lambda:1,alpha:0},{objective:"multi:softprob",eta:.01,max_depth:10,numRound:500,subsample:.7,colsample_bytree:.65,min_child_weight:.6,lambda:.1,alpha:0},{objective:"multi:softprob",eta:.1,max_depth:3,numRound:100,subsample:.9,colsample_bytree:.9,min_child_weight:1,lambda:1,alpha:0},{objective:"multi:softprob",eta:.03,max_depth:7,numRound:300,subsample:.8,colsample_bytree:.7,min_child_weight:1,lambda:5,alpha:1},{objective:"multi:softprob",eta:.02,max_depth:8,numRound:400,subsample:.8,colsample_bytree:.8,min_child_weight:.8,lambda:.1,alpha:0},{objective:"multi:softprob",eta:.08,max_depth:4,numRound:150,subsample:.85,colsample_bytree:.55,min_child_weight:1,lambda:1,alpha:.1},{objective:"multi:softprob",eta:.015,max_depth:9,numRound:350,subsample:.75,colsample_bytree:.55,min_child_weight:.9,lambda:.5,alpha:0},{objective:"multi:softprob",eta:.3,max_depth:3,numRound:50,subsample:.9,colsample_bytree:.9,min_child_weight:1,lambda:1,alpha:0},{objective:"multi:softprob",num_parallel_tree:100,numRound:1,subsample:.8,colsample_bynode:.8,learning_rate:1},{objective:"multi:softprob",num_parallel_tree:200,numRound:1,subsample:.7,colsample_bynode:.6,learning_rate:1}],lgb:[{objective:"multiclass",learning_rate:.05,max_depth:6,numRound:200,num_leaves:63,subsample:.8,colsample_bytree:.8,min_child_weight:1,reg_lambda:1,reg_alpha:0,verbosity:-1},{objective:"multiclass",learning_rate:.01,max_depth:-1,numRound:500,num_leaves:127,subsample:.7,colsample_bytree:.65,reg_lambda:.1,reg_alpha:0,verbosity:-1},{objective:"multiclass",learning_rate:.1,max_depth:4,numRound:100,num_leaves:15,subsample:.9,colsample_bytree:.9,reg_lambda:1,reg_alpha:0,verbosity:-1},{objective:"multiclass",learning_rate:.05,numRound:200,num_leaves:63,subsample:.8,colsample_bytree:.8,extra_trees:!0,reg_lambda:1,reg_alpha:0,verbosity:-1},{objective:"multiclass",learning_rate:.03,max_depth:7,numRound:300,num_leaves:63,subsample:.8,colsample_bytree:.7,reg_lambda:5,reg_alpha:1,verbosity:-1},{objective:"multiclass",learning_rate:.01,max_depth:8,numRound:500,num_leaves:95,subsample:.75,colsample_bytree:.55,reg_lambda:.5,reg_alpha:0,verbosity:-1}],ebm:[{objective:"classification",learningRate:.01,maxRounds:500,maxLeaves:3,maxBins:256},{objective:"classification",learningRate:.01,maxRounds:500,maxLeaves:4,maxInteractions:15,maxBins:256},{objective:"classification",learningRate:.05,maxRounds:300,maxLeaves:3,maxBins:128},{objective:"classification",learningRate:.005,maxRounds:800,maxLeaves:5,maxBins:512}],linear:[{solver:0,C:1},{solver:0,C:10},{solver:7,C:1},{solver:6,C:.1}],svm:[{svmType:0,kernel:2,C:1,gamma:0,probability:1},{svmType:0,kernel:2,C:10,gamma:.01,probability:1},{svmType:0,kernel:1,C:1,degree:3,gamma:0,probability:1},{svmType:0,kernel:0,C:1,probability:1}],knn:[{k:5,metric:"l2",task:"classification"},{k:15,metric:"l2",task:"classification"},{k:3,metric:"l1",task:"classification"}],tsetlin:[{task:"classification",nClauses:100,threshold:50,s:3,nEpochs:100},{task:"classification",nClauses:500,threshold:100,s:5,nEpochs:100},{task:"classification",nClauses:50,threshold:25,s:2,nEpochs:60}],rf:[{nEstimators:100,maxDepth:10,maxFeatures:"sqrt",criterion:"gini",minSamplesSplit:2,minSamplesLeaf:1},{nEstimators:300,maxDepth:0,maxFeatures:"log2",criterion:"gini",minSamplesSplit:5,minSamplesLeaf:2},{nEstimators:200,maxDepth:15,maxFeatures:"sqrt",criterion:"entropy",extraTrees:1,minSamplesSplit:2},{nEstimators:200,maxDepth:12,maxFeatures:"sqrt",criterion:"hellinger",minSamplesSplit:2,minSamplesLeaf:1},{nEstimators:150,maxDepth:10,maxFeatures:"sqrt",criterion:"gini",heterogeneous:1,oobWeighting:1,minSamplesLeaf:2},{nEstimators:200,maxDepth:10,maxFeatures:"sqrt",criterion:"gini",histogramBinning:1,minSamplesSplit:2,minSamplesLeaf:1},{nEstimators:150,maxDepth:12,maxFeatures:"sqrt",criterion:"gini",jarf:1,minSamplesSplit:2,minSamplesLeaf:1}],mlp:[{hidden_sizes:[64],activation:"relu",lr:.01,epochs:50,optimizer:"adam",batch_size:32},{hidden_sizes:[128,64],activation:"gelu",lr:.001,epochs:100,optimizer:"adam",batch_size:32},{hidden_sizes:[256,128],activation:"silu",lr:.001,epochs:150,optimizer:"adam",batch_size:16}],tabm:[{hidden_sizes:[128],activation:"relu",n_ensemble:32,lr:.005,epochs:100,optimizer:"adam"},{hidden_sizes:[64],activation:"gelu",n_ensemble:8,lr:.01,epochs:50,optimizer:"adam"},{hidden_sizes:[128,64],activation:"silu",n_ensemble:16,lr:.001,epochs:150,optimizer:"adam"}],nam:[{hidden_sizes:[64],activation:"exu",lr:.01,epochs:100,optimizer:"adam"},{hidden_sizes:[128],activation:"relu",lr:.001,epochs:100,optimizer:"adam"},{hidden_sizes:[64,32],activation:"exu",lr:.005,epochs:150,optimizer:"adam"}],gam:[{family:"binomial",penalty:"elasticnet",alpha:.5,nLambda:100,nFolds:5},{family:"binomial",penalty:"lasso",alpha:1,nLambda:100,nFolds:5},{family:"binomial",penalty:"ridge",alpha:0,nLambda:100,nFolds:5},{family:"binomial",penalty:"scad",nLambda:100,nFolds:5}],bart:[{numTrees:200,numBurnin:200,numSamples:100,alpha:.95,beta:2},{numTrees:500,numBurnin:300,numSamples:100,alpha:.95,beta:2},{numTrees:100,maxDepth:5,numBurnin:100,numSamples:100,alpha:.8,beta:1}],fm:[{k:4,lr:.01,epoch:10,opt:"adagrad"},{k:8,lr:.005,epoch:20,opt:"adagrad",lambda:.001}],xlr:[{lr:.01,opt:"ftrl",epoch:10},{lr:.01,lambda:.1,opt:"adagrad",epoch:20}]},regression:{xgb:[{objective:"reg:squarederror",eta:.05,max_depth:6,numRound:200,subsample:.8,colsample_bytree:.8,min_child_weight:1,lambda:1,alpha:0},{objective:"reg:squarederror",eta:.01,max_depth:10,numRound:500,subsample:.7,colsample_bytree:.65,min_child_weight:.6,lambda:.1,alpha:0},{objective:"reg:squarederror",eta:.1,max_depth:3,numRound:100,subsample:.9,colsample_bytree:.9,min_child_weight:1,lambda:1,alpha:0},{objective:"reg:squarederror",eta:.03,max_depth:7,numRound:300,subsample:.8,colsample_bytree:.7,min_child_weight:1,lambda:5,alpha:1},{objective:"reg:squarederror",eta:.02,max_depth:8,numRound:400,subsample:.8,colsample_bytree:.8,min_child_weight:.8,lambda:.1,alpha:0},{objective:"reg:squarederror",eta:.08,max_depth:4,numRound:150,subsample:.85,colsample_bytree:.55,min_child_weight:1,lambda:1,alpha:.1},{objective:"reg:squarederror",eta:.015,max_depth:9,numRound:350,subsample:.75,colsample_bytree:.55,min_child_weight:.9,lambda:.5,alpha:0},{objective:"reg:squarederror",eta:.3,max_depth:3,numRound:50,subsample:.9,colsample_bytree:.9,min_child_weight:1,lambda:1,alpha:0},{objective:"reg:squarederror",num_parallel_tree:100,numRound:1,subsample:.8,colsample_bynode:.8,learning_rate:1},{objective:"reg:squarederror",num_parallel_tree:200,numRound:1,subsample:.7,colsample_bynode:.6,learning_rate:1}],lgb:[{objective:"regression",learning_rate:.05,max_depth:6,numRound:200,num_leaves:63,subsample:.8,colsample_bytree:.8,min_child_weight:1,reg_lambda:1,reg_alpha:0,verbosity:-1},{objective:"regression",learning_rate:.01,max_depth:-1,numRound:500,num_leaves:127,subsample:.7,colsample_bytree:.65,reg_lambda:.1,reg_alpha:0,verbosity:-1},{objective:"regression",learning_rate:.1,max_depth:4,numRound:100,num_leaves:15,subsample:.9,colsample_bytree:.9,reg_lambda:1,reg_alpha:0,verbosity:-1},{objective:"regression",learning_rate:.05,numRound:200,num_leaves:63,subsample:.8,colsample_bytree:.8,extra_trees:!0,reg_lambda:1,reg_alpha:0,verbosity:-1},{objective:"regression",learning_rate:.03,max_depth:7,numRound:300,num_leaves:63,subsample:.8,colsample_bytree:.7,reg_lambda:5,reg_alpha:1,verbosity:-1},{objective:"regression",learning_rate:.01,max_depth:8,numRound:500,num_leaves:95,subsample:.75,colsample_bytree:.55,reg_lambda:.5,reg_alpha:0,verbosity:-1}],ebm:[{objective:"regression",learningRate:.01,maxRounds:500,maxLeaves:3,maxBins:256},{objective:"regression",learningRate:.01,maxRounds:500,maxLeaves:4,maxInteractions:15,maxBins:256},{objective:"regression",learningRate:.05,maxRounds:300,maxLeaves:3,maxBins:128},{objective:"regression",learningRate:.005,maxRounds:800,maxLeaves:5,maxBins:512}],linear:[{solver:11,C:1},{solver:11,C:10},{solver:12,C:1},{solver:13,C:.1}],svm:[{svmType:3,kernel:2,C:1,gamma:0},{svmType:3,kernel:2,C:10,gamma:.01},{svmType:3,kernel:1,C:1,degree:3,gamma:0},{svmType:3,kernel:0,C:1}],knn:[{k:5,metric:"l2",task:"regression"},{k:15,metric:"l2",task:"regression"},{k:3,metric:"l1",task:"regression"}],tsetlin:[{task:"regression",nClauses:100,threshold:50,s:3,nEpochs:100},{task:"regression",nClauses:500,threshold:100,s:5,nEpochs:100},{task:"regression",nClauses:50,threshold:25,s:2,nEpochs:60}],rf:[{nEstimators:100,maxDepth:10,maxFeatures:"sqrt",criterion:"mse",minSamplesSplit:2,minSamplesLeaf:1},{nEstimators:300,maxDepth:0,maxFeatures:"log2",criterion:"mse",minSamplesSplit:5,minSamplesLeaf:2},{nEstimators:200,maxDepth:15,maxFeatures:"sqrt",criterion:"mae",extraTrees:1,minSamplesSplit:2},{nEstimators:100,maxDepth:10,maxFeatures:"sqrt",criterion:"mse",leafModel:1,minSamplesLeaf:5},{nEstimators:150,maxDepth:10,maxFeatures:"sqrt",criterion:"mse",heterogeneous:1,oobWeighting:1,minSamplesLeaf:2},{nEstimators:200,maxDepth:10,maxFeatures:"sqrt",criterion:"mse",histogramBinning:1,minSamplesSplit:2,minSamplesLeaf:1},{nEstimators:150,maxDepth:12,maxFeatures:"sqrt",criterion:"mse",jarf:1,minSamplesSplit:2,minSamplesLeaf:1}],mlp:[{hidden_sizes:[64],activation:"relu",lr:.01,epochs:50,optimizer:"adam",batch_size:32},{hidden_sizes:[128,64],activation:"gelu",lr:.001,epochs:100,optimizer:"adam",batch_size:32},{hidden_sizes:[256,128],activation:"silu",lr:.001,epochs:150,optimizer:"adam",batch_size:16}],tabm:[{hidden_sizes:[128],activation:"relu",n_ensemble:32,lr:.005,epochs:100,optimizer:"adam"},{hidden_sizes:[64],activation:"gelu",n_ensemble:8,lr:.01,epochs:50,optimizer:"adam"},{hidden_sizes:[128,64],activation:"silu",n_ensemble:16,lr:.001,epochs:150,optimizer:"adam"}],nam:[{hidden_sizes:[64],activation:"exu",lr:.01,epochs:100,optimizer:"adam"},{hidden_sizes:[128],activation:"relu",lr:.001,epochs:100,optimizer:"adam"},{hidden_sizes:[64,32],activation:"exu",lr:.005,epochs:150,optimizer:"adam"}],gam:[{family:"gaussian",penalty:"elasticnet",alpha:.5,nLambda:100,nFolds:5},{family:"gaussian",penalty:"lasso",alpha:1,nLambda:100,nFolds:5},{family:"gaussian",penalty:"ridge",alpha:0,nLambda:100,nFolds:5},{family:"gaussian",penalty:"scad",nLambda:100,nFolds:5}],bart:[{numTrees:200,numBurnin:200,numSamples:100,alpha:.95,beta:2},{numTrees:500,numBurnin:300,numSamples:100,alpha:.95,beta:2},{numTrees:100,maxDepth:5,numBurnin:100,numSamples:100,alpha:.8,beta:1}],fm:[{k:4,lr:.01,epoch:10,opt:"adagrad"},{k:8,lr:.005,epoch:20,opt:"adagrad",lambda:.001}],xlr:[{lr:.01,opt:"ftrl",epoch:10},{lr:.01,lambda:.1,opt:"adagrad",epoch:20}]}};function Ts(s="classification"){return ne[s]||ne.classification}var wt=class{#t=[];#e=0;#s=0;constructor(t,{task:n="classification",seed:e=42}={}){let i=Ts(n);for(let r of t){let o=r.name,a=r.cls,l=r.params||{},h=i[o]||[{}];for(let c of h){let f={...c,...l},u=_o(o,f);this.#t.push({candidateId:u,cls:a,params:f})}}this.#s=this.#t.length}next(){return this.#e>=this.#s?null:this.#t[this.#e++]}report(t){}isDone(){return this.#e>=this.#s}},re=class{#t;#e;#s=null;#n=null;constructor(t,n={}){if(!t||t.length===0)throw new se("PortfolioSearch: at least one model is required");this.#t=t,this.#e={scoring:null,cv:5,seed:42,task:null,maxTimeMs:0,onProgress:null,...n}}async fit(t,n){let e=Cs(t),i=js(n),r=this.#e.task||So(i),o=this.#e.scoring||(r==="classification"?"accuracy":"r2"),{cv:a,seed:l,maxTimeMs:h,onProgress:c}=this.#e,f=r==="classification"?wo(i,a,{shuffle:!0,seed:l}):yo(i.length,a,{shuffle:!0,seed:l}),u=new xo({folds:f,scoring:o,X:e,y:i,timeLimitMs:h,seed:l,onProgress:c}),d=new wt(this.#t,{task:r,seed:l}),{leaderboard:m}=await u.runStrategy(d);if(m.length===0)throw new se("PortfolioSearch: no candidates were evaluated");return this.#s=m,this.#n=m.best(),{leaderboard:m,bestResult:this.#n}}async refitBest(t,n){if(!this.#n)throw new se("PortfolioSearch: must call fit() first");let e=this.#n,r=await this.#t.find(l=>l.name===e.modelName).cls.create(e.params),o=Cs(t),a=js(n);return r.fit(o,a),r}get leaderboard(){return this.#s}get bestResult(){return this.#n}};$s.exports={PORTFOLIO:ne,getPortfolio:Ts,PortfolioStrategy:wt,PortfolioSearch:re}});var Ks=S((Cc,Js)=>{var{encodeBundle:vo,decodeBundle:Fo,register:Ds,load:Xs,normalizeX:oe,normalizeY:Vs,accuracy:Ao,r2Score:ko,ValidationError:yt,NotFittedError:qo,DisposedError:Eo,lift:Us}=q(),Ys="wlearn.ensemble.voting.classifier@1",Gs="wlearn.ensemble.voting.regressor@1",Ws=!1,ae=class s{#t;#e;#s;#n;#i;#r;#o=!1;#l=!1;constructor(t){this.#t=t.estimators||[],this.#s=t.weights||null,this.#n=t.voting||"soft",this.#i=t.task||"classification",this.#e=null,this.#r=null,s._register()}static async create(t={}){return new s(t)}#a(){if(this.#l)throw new Eo("VotingEnsemble has been disposed.")}#c(){if(this.#a(),!this.#o)throw new qo("VotingEnsemble is not fitted. Call fit() first.")}async fit(t,n){this.#a();let e=oe(t),i=Vs(n);if(this.#i==="classification"){let r=new Set;for(let o=0;o<i.length;o++)r.add(i[o]);this.#r=new Int32Array([...r].sort((o,a)=>o-a))}this.#s||(this.#s=new Float64Array(this.#t.length).fill(1/this.#t.length)),this.#e=[];for(let[r,o,a]of this.#t){let l=await o.create(a||{});l.fit(e,i),this.#e.push(l)}return this.#o=!0,this}predict(t){this.#c();let n=oe(t),e=n.rows;if(this.#i==="regression")return this.#f(n,e);if(this.#n==="soft"){let i=this.predictProba(n);return Us(i,r=>{let o=this.#r.length,a=new Float64Array(e);for(let l=0;l<e;l++){let h=0,c=-1/0;for(let f=0;f<o;f++)r[l*o+f]>c&&(c=r[l*o+f],h=f);a[l]=this.#r[h]}return a})}return this.#h(n,e)}predictProba(t){if(this.#c(),this.#i!=="classification")throw new yt("predictProba is only available for classification");if(this.#n==="hard")throw new yt('predictProba requires voting="soft"');let n=oe(t),e=n.rows,i=this.#r.length,r=[],o=!1;for(let l=0;l<this.#e.length;l++){let h=this.#e[l].predictProba(n);h!=null&&typeof h.then=="function"&&(o=!0),r.push(h)}let a=l=>{let h=new Float64Array(e*i);for(let c=0;c<l.length;c++){let f=l[c],u=this.#s[c];for(let d=0;d<e*i;d++)h[d]+=u*f[d]}return h};return o?Promise.all(r).then(a):a(r)}score(t,n){this.#c();let e=this.predict(t),i=Vs(n),r=this.#i==="classification"?Ao:ko;return Us(e,o=>r(i,o))}save(){this.#c();let n={typeId:this.#i==="classification"?Ys:Gs,params:{task:this.#i,voting:this.#n,weights:[...this.#s],estimatorNames:this.#t.map(i=>i[0]),classes:this.#r?[...this.#r]:null}},e=this.#e.map((i,r)=>({id:this.#t[r][0],data:i.save(),mediaType:"application/x-wlearn-bundle"}));return vo(n,e)}static async load(t){let{manifest:n,toc:e,blobs:i}=Fo(t),r=n.params,o=new s({task:r.task,voting:r.voting,weights:new Float64Array(r.weights)});o.#r=r.classes?new Int32Array(r.classes):null,o.#t=r.estimatorNames.map(a=>[a,null,null]),o.#e=[];for(let a of r.estimatorNames){let l=e.find(f=>f.id===a);if(!l)throw new yt(`No artifact for estimator "${a}"`);let h=i.subarray(l.offset,l.offset+l.length),c=await Xs(h);o.#e.push(c)}return o.#o=!0,o}dispose(){if(!this.#l&&(this.#l=!0,this.#e))for(let t of this.#e)t.dispose()}getParams(){return{task:this.#i,voting:this.#n,weights:this.#s?[...this.#s]:null,estimatorNames:this.#t.map(t=>t[0])}}setParams(t){return this.#a(),t.voting!==void 0&&(this.#n=t.voting),t.weights!==void 0&&(this.#s=new Float64Array(t.weights)),this}get capabilities(){return{classifier:this.#i==="classification",regressor:this.#i==="regression",predictProba:this.#i==="classification"&&this.#n==="soft",decisionFunction:!1,sampleWeight:!1,csr:!1,earlyStopping:!1}}get isFitted(){return this.#o}get classes(){return this.#r}#f(t,n){let e=[],i=!1;for(let o=0;o<this.#e.length;o++){let a=this.#e[o].predict(t);a!=null&&typeof a.then=="function"&&(i=!0),e.push(a)}let r=o=>{let a=new Float64Array(n);for(let l=0;l<o.length;l++){let h=this.#s[l];for(let c=0;c<n;c++)a[c]+=h*o[l][c]}return a};return i?Promise.all(e).then(r):r(e)}#h(t,n){let e=[],i=!1;for(let o=0;o<this.#e.length;o++){let a=this.#e[o].predict(t);a!=null&&typeof a.then=="function"&&(i=!0),e.push(a)}let r=o=>{let a=this.#r.length,l=new Float64Array(n);for(let h=0;h<n;h++){let c=new Float64Array(a);for(let d=0;d<o.length;d++){let m=o[d][h],p=this.#r.indexOf(m);p>=0&&(c[p]+=this.#s[d])}let f=0,u=-1/0;for(let d=0;d<a;d++)c[d]>u&&(u=c[d],f=d);l[h]=this.#r[f]}return l};return i?Promise.all(e).then(r):r(e)}static _register(){if(Ws)return;Ws=!0;let t=(n,e,i)=>s._loadFromParts(n,e,i);Ds(Ys,t),Ds(Gs,t)}static async _loadFromParts(t,n,e){let i=t.params,r=new s({task:i.task,voting:i.voting,weights:new Float64Array(i.weights)});r.#r=i.classes?new Int32Array(i.classes):null,r.#t=i.estimatorNames.map(o=>[o,null,null]),r.#e=[];for(let o of i.estimatorNames){let a=n.find(c=>c.id===o);if(!a)throw new yt(`No artifact for estimator "${o}"`);let l=e.subarray(a.offset,a.offset+a.length),h=await Xs(l);r.#e.push(h)}return r.#o=!0,r}};Js.exports={VotingEnsemble:ae}});var an=S((jc,on)=>{var{encodeBundle:Io,decodeBundle:Po,register:Hs,load:Zs,normalizeX:Qs,normalizeY:tn,accuracy:Ro,r2Score:Mo,stratifiedKFold:zo,kFold:No,ValidationError:rt,NotFittedError:Lo,DisposedError:Bo,lift:ce}=q(),en="wlearn.ensemble.stacking.classifier@1",sn="wlearn.ensemble.stacking.regressor@1",nn=!1,le=class s{#t;#e;#s;#n;#i;#r;#o;#l;#a;#c;#f;#h=!1;#d=!1;constructor(t){this.#t=t.estimators||[],this.#e=t.finalEstimator||null,this.#i=t.cv||5,this.#r=t.task||"classification",this.#o=t.passthrough||!1,this.#l=t.seed??42,this.#s=null,this.#n=null,this.#a=null,this.#c=0,this.#f=0,s._register()}static async create(t={}){return new s(t)}#m(){if(this.#d)throw new Bo("StackingEnsemble has been disposed.")}#u(){if(this.#m(),!this.#h)throw new Lo("StackingEnsemble is not fitted. Call fit() first.")}async fit(t,n){if(this.#m(),!this.#e)throw new rt("StackingEnsemble requires a finalEstimator");let e=Qs(t),i=tn(n),r=e.rows;if(this.#r==="classification"){let m=new Set;for(let p=0;p<i.length;p++)m.add(i[p]);this.#a=new Int32Array([...m].sort((p,g)=>p-g)),this.#c=this.#a.length}let o=this.#r==="classification"?zo(i,this.#i,{shuffle:!0,seed:this.#l}):No(r,this.#i,{shuffle:!0,seed:this.#l}),a=this.#t.length,l=this.#r==="classification"?this.#c:1,h=a*l,c=new Float64Array(r*h);for(let m=0;m<a;m++){let[,p,g]=this.#t[m];for(let{train:w,test:b}of o){let y=rn(e,w),_=Oo(i,w),A=rn(e,b),v=await p.create(g||{});try{if(v.fit(y,_),this.#r==="classification"){let F=await v.predictProba(A);for(let k=0;k<b.length;k++){let O=b[k];for(let P=0;P<this.#c;P++)c[O*h+m*l+P]=F[k*this.#c+P]}}else{let F=await v.predict(A);for(let k=0;k<b.length;k++)c[b[k]*h+m]=F[k]}}finally{v.dispose()}}}let f;if(this.#o){this.#f=h+e.cols;let m=new Float64Array(r*this.#f);for(let p=0;p<r;p++)m.set(c.subarray(p*h,(p+1)*h),p*this.#f),m.set(e.data.subarray(p*e.cols,(p+1)*e.cols),p*this.#f+h);f={data:m,rows:r,cols:this.#f}}else this.#f=h,f={data:c,rows:r,cols:h};this.#s=[];for(let[,m,p]of this.#t){let g=await m.create(p||{});g.fit(e,i),this.#s.push(g)}let[,u,d]=this.#e;return this.#n=await u.create(d||{}),this.#n.fit(f,i),this.#h=!0,this}predict(t){this.#u();let n=this.#p(t);return ce(n,e=>this.#n.predict(e))}predictProba(t){if(this.#u(),this.#r!=="classification")throw new rt("predictProba is only available for classification");if(typeof this.#n.predictProba!="function")throw new rt("Meta-model does not support predictProba");let n=this.#p(t);return ce(n,e=>this.#n.predictProba(e))}score(t,n){this.#u();let e=this.predict(t),i=tn(n),r=this.#r==="classification"?Ro:Mo;return ce(e,o=>r(i,o))}save(){this.#u();let n={typeId:this.#r==="classification"?en:sn,params:{task:this.#r,cv:this.#i,passthrough:this.#o,seed:this.#l,estimatorNames:this.#t.map(i=>i[0]),metaName:this.#e[0],classes:this.#a?[...this.#a]:null,nMetaCols:this.#f}},e=this.#s.map((i,r)=>({id:this.#t[r][0],data:i.save(),mediaType:"application/x-wlearn-bundle"}));return e.push({id:this.#e[0],data:this.#n.save(),mediaType:"application/x-wlearn-bundle"}),Io(n,e)}static async load(t){let{manifest:n,toc:e,blobs:i}=Po(t);return s._loadFromParts(n,e,i)}dispose(){if(!this.#d){if(this.#d=!0,this.#s)for(let t of this.#s)t.dispose();this.#n&&this.#n.dispose()}}getParams(){return{task:this.#r,cv:this.#i,passthrough:this.#o,seed:this.#l,estimatorNames:this.#t.map(t=>t[0]),metaName:this.#e?this.#e[0]:null}}setParams(t){return this.#m(),t.cv!==void 0&&(this.#i=t.cv),t.passthrough!==void 0&&(this.#o=t.passthrough),t.seed!==void 0&&(this.#l=t.seed),this}get capabilities(){return{classifier:this.#r==="classification",regressor:this.#r==="regression",predictProba:this.#r==="classification",decisionFunction:!1,sampleWeight:!1,csr:!1,earlyStopping:!1}}get isFitted(){return this.#h}get classes(){return this.#a}#p(t){let n=Qs(t),e=n.rows,i=this.#s.length,r=this.#r==="classification"?this.#c:1,o=i*r,a=[],l=!1;for(let c=0;c<i;c++){let f=this.#r==="classification"?this.#s[c].predictProba(n):this.#s[c].predict(n);f!=null&&typeof f.then=="function"&&(l=!0),a.push(f)}let h=c=>{let f=new Float64Array(e*this.#f);for(let u=0;u<i;u++)if(this.#r==="classification"){let d=c[u];for(let m=0;m<e;m++)for(let p=0;p<this.#c;p++)f[m*this.#f+u*r+p]=d[m*this.#c+p]}else{let d=c[u];for(let m=0;m<e;m++)f[m*this.#f+u]=d[m]}if(this.#o)for(let u=0;u<e;u++)f.set(n.data.subarray(u*n.cols,(u+1)*n.cols),u*this.#f+o);return{data:f,rows:e,cols:this.#f}};return l?Promise.all(a).then(h):h(a)}static _register(){if(nn)return;nn=!0;let t=(n,e,i)=>s._loadFromParts(n,e,i);Hs(en,t),Hs(sn,t)}static async _loadFromParts(t,n,e){let i=t.params,r=new s({task:i.task,cv:i.cv,passthrough:i.passthrough,seed:i.seed});r.#a=i.classes?new Int32Array(i.classes):null,r.#c=r.#a?r.#a.length:0,r.#f=i.nMetaCols,r.#t=i.estimatorNames.map(l=>[l,null,null]),r.#e=[i.metaName,null,null],r.#s=[];for(let l of i.estimatorNames){let h=n.find(f=>f.id===l);if(!h)throw new rt(`No artifact for base estimator "${l}"`);let c=e.subarray(h.offset,h.offset+h.length);r.#s.push(await Zs(c))}let o=n.find(l=>l.id===i.metaName);if(!o)throw new rt(`No artifact for meta estimator "${i.metaName}"`);let a=e.subarray(o.offset,o.offset+o.length);return r.#n=await Zs(a),r.#h=!0,r}};function rn(s,t){let{data:n,cols:e}=s,i=t.length,r=new Float64Array(i*e);for(let o=0;o<i;o++){let a=t[o]*e;r.set(n.subarray(a,a+e),o*e)}return{data:r,rows:i,cols:e}}function Oo(s,t){let n=new s.constructor(t.length);for(let e=0;e<t.length;e++)n[e]=s[t[e]];return n}on.exports={StackingEnsemble:le}});var bn=S((Tc,gn)=>{var{encodeBundle:Co,decodeBundle:jo,register:cn,load:To,normalizeX:fe,normalizeY:ln,accuracy:$o,r2Score:Do,stratifiedKFold:Xo,kFold:Vo,ValidationError:fn,NotFittedError:Uo,DisposedError:Yo,lift:hn}=q(),un="wlearn.ensemble.bagged.classifier@1",dn="wlearn.ensemble.bagged.regressor@1",mn=!1,he=class s{#t;#e;#s;#n;#i;#r;#o;#l=0;#a=0;#c;#f;#h=!1;#d=!1;constructor(t={}){this.#t=t.estimator||null,this.#e=t.kFold||5,this.#s=t.nRepeats||1,this.#n=t.task||"classification",this.#i=t.seed??42,this.#r=null,this.#o=null,this.#c=null,this.#f=null,s._register()}static async create(t={}){return new s(t)}#m(){if(this.#d)throw new Yo("BaggedEstimator has been disposed.")}#u(){if(this.#m(),!this.#h)throw new Uo("BaggedEstimator is not fitted. Call fit() first.")}async fit(t,n){this.#m();let e=fe(t),i=ln(n),r=e.rows;if(this.#a=r,this.#n==="classification"){let l=new Set;for(let h=0;h<i.length;h++)l.add(i[h]);this.#o=new Int32Array([...l].sort((h,c)=>h-c)),this.#l=this.#o.length}this.#n==="classification"?this.#c=new Float64Array(r*this.#l):this.#c=new Float64Array(r),this.#f=new Uint8Array(r);let[,o,a]=this.#t;this.#r=[];for(let l=0;l<this.#s;l++){let h=this.#i+l,c=this.#n==="classification"?Xo(i,this.#e,{shuffle:!0,seed:h}):Vo(r,this.#e,{shuffle:!0,seed:h});for(let{train:f,test:u}of c){let d=pn(e,f),m=Go(i,f),p=pn(e,u),g=await o.create(a||{});if(g.fit(d,m),this.#n==="classification"){let w=await g.predictProba(p),b=this.#l;for(let y=0;y<u.length;y++){let _=u[y];for(let A=0;A<b;A++)this.#c[_*b+A]+=w[y*b+A]}}else{let w=await g.predict(p);for(let b=0;b<u.length;b++)this.#c[u[b]]+=w[b]}for(let w=0;w<u.length;w++)this.#f[u[w]]+=1;this.#r.push(g)}}return this.#h=!0,this}predict(t){this.#u();let n=fe(t),e=n.rows;if(this.#n==="regression")return this.#p(n,e);let i=this.predictProba(n);return hn(i,r=>{let o=this.#l,a=new Float64Array(e);for(let l=0;l<e;l++){let h=0,c=-1/0;for(let f=0;f<o;f++)r[l*o+f]>c&&(c=r[l*o+f],h=f);a[l]=this.#o[h]}return a})}predictProba(t){if(this.#u(),this.#n!=="classification")throw new fn("predictProba is only available for classification");let n=fe(t),e=n.rows,i=this.#l,r=this.#r.length,o=[],a=!1;for(let h of this.#r){let c=h.predictProba(n);c!=null&&typeof c.then=="function"&&(a=!0),o.push(c)}let l=h=>{let c=new Float64Array(e*i);for(let f of h)for(let u=0;u<e*i;u++)c[u]+=f[u];for(let f=0;f<e*i;f++)c[f]/=r;return c};return a?Promise.all(o).then(l):l(o)}score(t,n){this.#u();let e=this.predict(t),i=ln(n),r=this.#n==="classification"?$o:Do;return hn(e,o=>r(i,o))}#p(t,n){let e=this.#r.length,i=[],r=!1;for(let a of this.#r){let l=a.predict(t);l!=null&&typeof l.then=="function"&&(r=!0),i.push(l)}let o=a=>{let l=new Float64Array(n);for(let h of a)for(let c=0;c<n;c++)l[c]+=h[c];for(let h=0;h<n;h++)l[h]/=e;return l};return r?Promise.all(i).then(o):o(i)}get oofPredictions(){this.#u();let t=new Uint8Array(this.#f);for(let e=0;e<t.length;e++)t[e]===0&&(t[e]=1);if(this.#n==="classification"){let e=this.#l,i=new Float64Array(this.#c);for(let r=0;r<this.#a;r++){let o=t[r];for(let a=0;a<e;a++)i[r*e+a]/=o}return i}let n=new Float64Array(this.#c);for(let e=0;e<this.#a;e++)n[e]/=t[e];return n}save(){this.#u();let n={typeId:this.#n==="classification"?un:dn,params:{task:this.#n,kFold:this.#e,nRepeats:this.#s,seed:this.#i,estimatorName:this.#t[0],classes:this.#o?[...this.#o]:null,nClasses:this.#l,nSamples:this.#a}},e=this.#r.map((o,a)=>({id:`fold_${a}`,data:o.save(),mediaType:"application/x-wlearn-bundle"})),i=this.oofPredictions,r=new Uint8Array(i.buffer,i.byteOffset,i.byteLength);return e.push({id:"oof",data:r,mediaType:"application/octet-stream"}),Co(n,e)}static async load(t){let{manifest:n,toc:e,blobs:i}=jo(t);return s._loadFromParts(n,e,i)}dispose(){if(!this.#d){if(this.#d=!0,this.#r)for(let t of this.#r)t.dispose();this.#r=null,this.#c=null,this.#f=null}}getParams(){return{task:this.#n,kFold:this.#e,nRepeats:this.#s,seed:this.#i,estimatorName:this.#t?this.#t[0]:null}}setParams(t){return this.#m(),t.kFold!==void 0&&(this.#e=t.kFold),t.nRepeats!==void 0&&(this.#s=t.nRepeats),t.seed!==void 0&&(this.#i=t.seed),this}get capabilities(){return{classifier:this.#n==="classification",regressor:this.#n==="regression",predictProba:this.#n==="classification",decisionFunction:!1,sampleWeight:!1,csr:!1,earlyStopping:!1}}get isFitted(){return this.#h&&!this.#d}get classes(){return this.#o}static _register(){if(mn)return;mn=!0;let t=(n,e,i)=>s._loadFromParts(n,e,i);cn(un,t),cn(dn,t)}static async _loadFromParts(t,n,e){let i=t.params,r=new s({task:i.task,kFold:i.kFold||5,nRepeats:i.nRepeats||1,seed:i.seed??42});r.#o=i.classes?new Int32Array(i.classes):null,r.#l=i.nClasses||0,r.#a=i.nSamples||0,r.#t=[i.estimatorName||"base",null,null];let o=r.#e*r.#s;r.#r=[];for(let l=0;l<o;l++){let h=`fold_${l}`,c=n.find(u=>u.id===h);if(!c)throw new fn(`No artifact for "${h}"`);let f=e.subarray(c.offset,c.offset+c.length);r.#r.push(await To(f))}let a=n.find(l=>l.id==="oof");if(a){let l=e.subarray(a.offset,a.offset+a.length),h=new Float64Array(l.buffer.slice(l.byteOffset,l.byteOffset+l.byteLength));r.#c=h,r.#f=new Uint8Array(r.#a).fill(1)}else r.#n==="classification"?r.#c=new Float64Array(r.#a*r.#l):r.#c=new Float64Array(r.#a),r.#f=new Uint8Array(r.#a);return r.#h=!0,r}};function pn(s,t){let{data:n,cols:e}=s,i=t.length,r=new Float64Array(i*e);for(let o=0;o<i;o++){let a=t[o]*e;r.set(n.subarray(a,a+e),o*e)}return{data:r,rows:i,cols:e}}function Go(s,t){let n=new s.constructor(t.length);for(let e=0;e<t.length;e++)n[e]=s[t[e]];return n}gn.exports={BaggedEstimator:he}});var de=S(($c,yn)=>{var{ValidationError:wn}=q();function ue(s){let t=s.length;if(t===0)return new Float64Array(0);if(t===1)return new Float64Array([1]);let n=new Float64Array(s);n.sort(),n.reverse();let e=new Float64Array(t);e[0]=n[0];for(let a=1;a<t;a++)e[a]=e[a-1]+n[a];let i=-1;for(let a=0;a<t;a++)n[a]*(a+1)>e[a]-1&&(i=a);if(i<0){let a=new Float64Array(t);return a.fill(1/t),a}let r=(e[i]-1)/(i+1),o=new Float64Array(t);for(let a=0;a<t;a++)o[a]=Math.max(s[a]-r,0);return o}function Wo(s,t,n,{task:e="classification",lr:i=.05,nIter:r=100}={}){let o=s.length,a=t.length;if(o===0)throw new wn("optimizeWeights: need at least 1 model");if(o===1)return new Float64Array([1]);let l=new Float64Array(n),h=1e-15;if(e==="classification"){let f=s[0].length/a;if(f!==Math.floor(f))throw new wn("optimizeWeights: prediction length must be divisible by n");for(let u=0;u<r;u++){let d=new Float64Array(o);for(let p=0;p<a;p++){let g=t[p]|0,w=0;for(let b=0;b<o;b++)w+=l[b]*s[b][p*f+g];w=Math.max(w,h);for(let b=0;b<o;b++)d[b]-=s[b][p*f+g]/w}for(let p=0;p<o;p++)d[p]/=a;for(let p=0;p<o;p++)l[p]-=i*d[p];let m=ue(l);for(let p=0;p<o;p++)l[p]=m[p]}}else for(let c=0;c<r;c++){let f=new Float64Array(o);for(let d=0;d<a;d++){let m=0;for(let g=0;g<o;g++)m+=l[g]*s[g][d];let p=t[d]-m;for(let g=0;g<o;g++)f[g]-=2*p*s[g][d]}for(let d=0;d<o;d++)f[d]/=a;for(let d=0;d<o;d++)l[d]-=i*f[d];let u=ue(l);for(let d=0;d<o;d++)l[d]=u[d]}return l}yn.exports={optimizeWeights:Wo,projectSimplex:ue}});var _n=S((Dc,Sn)=>{var{getScorer:Jo,normalizeY:Ko,ValidationError:xn}=q(),{optimizeWeights:Ho}=de();function Zo(s,t,{maxSize:n=20,scoring:e="accuracy",task:i="classification",nClasses:r=0,refineWeights:o=!0}={}){let a=Ko(t),l=a.length,h=s.length;if(h===0)throw new xn("caruanaSelect: need at least 1 candidate");let c=Jo(e),f=s[0].length/l;if(f!==Math.floor(f))throw new xn("caruanaSelect: oofPredictions[0].length must be divisible by n");i==="classification"&&r===0&&(r=f);let u=new Float64Array(s[0].length),d=[],m=[];for(let y=0;y<n;y++){let _=-1,A=-1/0;for(let F=0;F<h;F++){let k=Qo(u,s[F],y,y+1),O=ta(k,a,c,i,r,l);O>A&&(A=O,_=F)}d.push(_),m.push(A);let v=s[_];for(let F=0;F<u.length;F++)u[F]=(y*u[F]+v[F])/(y+1)}let p=new Map;for(let y of d)p.set(y,(p.get(y)||0)+1);let g=new Int32Array([...p.keys()].sort((y,_)=>y-_)),w=new Float64Array(g.length);for(let y=0;y<g.length;y++)w[y]=p.get(g[y])/n;let b={indices:g,weights:w,scores:new Float64Array(m)};if(o&&g.length>1){let y=Array.from(g,_=>s[_]);b.weights=Ho(y,a,w,{task:i})}return b}function Qo(s,t,n,e){let i=new Float64Array(s.length);for(let r=0;r<s.length;r++)i[r]=(n*s[r]+t[r])/e;return i}function ta(s,t,n,e,i,r){if(e==="regression")return n(t,s);let o=new Float64Array(r);for(let a=0;a<r;a++){let l=0,h=-1/0;for(let c=0;c<i;c++)s[a*i+c]>h&&(h=s[a*i+c],l=c);o[a]=l}return n(t,o)}Sn.exports={caruanaSelect:Zo}});var An=S((Vc,Fn)=>{var{stratifiedKFold:ea,kFold:sa,normalizeX:na,normalizeY:ra,ValidationError:Xc}=q();async function ia(s,t,n,{cv:e=5,seed:i=42,task:r="classification"}={}){let o=na(t),a=ra(n),l=o.rows,h=r==="classification"?ea(a,e,{shuffle:!0,seed:i}):sa(l,e,{shuffle:!0,seed:i}),c=null,f=0;if(r==="classification"){let d=new Set;for(let m=0;m<a.length;m++)d.add(a[m]);c=new Int32Array([...d].sort((m,p)=>m-p)),f=c.length}let u=[];for(let[d,m,p]of s){let g;r==="classification"?g=new Float64Array(l*f):g=new Float64Array(l);for(let{train:w,test:b}of h){let y=vn(o,w),_=oa(a,w),A=vn(o,b),v=await m.create(p||{});try{if(v.fit(y,_),r==="classification"){let F=await v.predictProba(A);for(let k=0;k<b.length;k++){let O=b[k];for(let P=0;P<f;P++)g[O*f+P]=F[k*f+P]}}else{let F=await v.predict(A);for(let k=0;k<b.length;k++)g[b[k]]=F[k]}}finally{v.dispose()}}u.push(g)}return{oofPreds:u,classes:c}}function vn(s,t){let{data:n,cols:e}=s,i=t.length,r=new Float64Array(i*e);for(let o=0;o<i;o++){let a=t[o]*e;r.set(n.subarray(a,a+e),o*e)}return{data:r,rows:i,cols:e}}function oa(s,t){let n=new s.constructor(t.length);for(let e=0;e<t.length;e++)n[e]=s[t[e]];return n}Fn.exports={getOofPredictions:ia}});var qn=S((Uc,kn)=>{var{VotingEnsemble:aa}=Ks(),{StackingEnsemble:ca}=an(),{BaggedEstimator:la}=bn(),{caruanaSelect:fa}=_n(),{getOofPredictions:ha}=An(),{optimizeWeights:ua,projectSimplex:da}=de();kn.exports={VotingEnsemble:aa,StackingEnsemble:ca,BaggedEstimator:la,caruanaSelect:fa,getOofPredictions:ha,optimizeWeights:ua,projectSimplex:da}});var pe=S((Gc,In)=>{var{makeLCG:En}=q(),{sampleConfig:ma}=st(),{makeCandidateId:pa}=M(),{max:ga,ceil:Yc}=Math,me=class{#t=[];#e=[];#s="probe";#n=0;#i=0;#r=[];#o;#l;#a=!1;#c;constructor(t,{nIter:n=20,seed:e=42,promoteCount:i=10,greaterIsBetter:r=!0,probeFraction:o=.5}={}){this.#o=i,this.#l=r,this.#c=o;let a=En(e);for(let l of t){let c={...l.searchSpace||l.cls.defaultSearchSpace?.()||{}};if(l.params)for(let u of Object.keys(l.params))delete c[u];let f=En(a()*2147483647|0);for(let u=0;u<n;u++){let m={...ma(c,f),...l.params||{}},p=pa(l.name,m);this.#t.push({candidateId:p,cls:l.cls,params:m})}}}get phase(){return this.#s}next(){if(this.#a)return null;if(this.#s==="probe"){if(this.#n>=this.#t.length)return null;let t=this.#t[this.#n++];return this.#c<1?{...t,budget:{type:"subsample",value:this.#c}}:t}return this.#i>=this.#e.length?null:this.#e[this.#i++]}report(t){if(this.#s==="probe"){this.#r.push(t),this.#r.length>=this.#t.length&&this.#f();return}}#f(){let t=[...this.#r];this.#l?t.sort((i,r)=>r.meanScore-i.meanScore):t.sort((i,r)=>i.meanScore-r.meanScore);let n=t.slice(0,ga(1,this.#o)),e=new Set(n.map(i=>i.candidateId));this.#e=this.#t.filter(i=>e.has(i.candidateId)),this.#s="promote",this.#i=0}isDone(){return this.#a?!0:this.#s==="promote"&&this.#i>=this.#e.length&&this.#e.length>0?(this.#a=!0,!0):!1}};In.exports={ProgressiveStrategy:me}});var we=S((Wc,Ln)=>{var{stratifiedKFold:Pn,kFold:Rn,normalizeX:Mn,normalizeY:zn,ValidationError:ge}=q(),{Executor:Nn}=H(),{ProgressiveStrategy:ba}=pe(),{detectTask:wa,scorerGreaterIsBetter:ya}=M(),be=class{#t;#e;#s=null;#n=null;constructor(t,n={}){if(!t||t.length===0)throw new ge("ProgressiveSearch: at least one model is required");this.#t=t,this.#e={scoring:null,cv:5,seed:42,task:null,nIter:20,maxTimeMs:0,promoteCount:10,probeFraction:.5,onProgress:null,...n}}async fit(t,n){let e=Mn(t),i=zn(n),r=this.#e.task||wa(i),o=this.#e.scoring||(r==="classification"?"accuracy":"r2"),{cv:a,seed:l,nIter:h,maxTimeMs:c,promoteCount:f,probeFraction:u,onProgress:d}=this.#e,m=ya(o),g=[(r==="classification"?Pn(i,2,{shuffle:!0,seed:l}):Rn(i.length,2,{shuffle:!0,seed:l}))[0]],w=r==="classification"?Pn(i,a,{shuffle:!0,seed:l+1}):Rn(i.length,a,{shuffle:!0,seed:l+1}),b=new ba(this.#t,{nIter:h,seed:l,promoteCount:f,greaterIsBetter:m,probeFraction:u}),y=new Nn({folds:g,scoring:o,X:e,y:i,timeLimitMs:c>0?Math.floor(c*.3):0,seed:l,onProgress:d});for(;b.phase==="probe"&&!b.isDone()&&!y.isTimedOut;){let v=b.next();if(v===null)break;try{let F=await y.evaluateCandidate(v);b.report(F)}catch{b.report({candidateId:v.candidateId,meanScore:-1/0,foldScores:new Float64Array(1),stdScore:0,fitTimeMs:0,nTrainUsed:0,nTest:0})}}b.phase;let _=new Nn({folds:w,scoring:o,X:e,y:i,timeLimitMs:c>0?Math.floor(c*.7):0,seed:l,onProgress:d});for(;!b.isDone()&&!_.isTimedOut;){let v=b.next();if(v===null)break;try{await _.evaluateCandidate(v)}catch{}}let A=_.leaderboard;if(A.length===0){let v=y.leaderboard;if(v.length===0)throw new ge("ProgressiveSearch: no candidates were evaluated");this.#s=v}else this.#s=A;return this.#n=this.#s.best(),{leaderboard:this.#s,bestResult:this.#n}}async refitBest(t,n){if(!this.#n)throw new ge("ProgressiveSearch: must call fit() first");let e=this.#n,r=await this.#t.find(l=>l.name===e.modelName).cls.create(e.params),o=Mn(t),a=zn(n);return r.fit(o,a),r}get leaderboard(){return this.#s}get bestResult(){return this.#n}};Ln.exports={ProgressiveSearch:be}});var Tn=S((Jc,jn)=>{var{normalizeX:Bn,normalizeY:On,ValidationError:Cn,Preprocessor:xa}=q(),{getOofPredictions:Sa,caruanaSelect:_a,VotingEnsemble:va,StackingEnsemble:Fa}=qn(),{RandomSearch:Aa}=Jt(),{SuccessiveHalvingSearch:ka}=ee(),{PortfolioSearch:qa}=ie(),{ProgressiveSearch:Ea}=we(),{detectTask:Ia}=M();function Pa(s,t,n,e){if(e==="classification"){let f=s.length/n,u=0;for(let d=0;d<n;d++){let m=0,p=0,g=-1/0,w=-1/0;for(let b=0;b<f;b++){let y=d*f+b;s[y]>g&&(g=s[y],m=b),t[y]>w&&(w=t[y],p=b)}m!==p&&u++}return u/n}let i=0,r=0,o=0,a=0,l=0;for(let f=0;f<n;f++)i+=s[f],r+=t[f],o+=s[f]*s[f],a+=t[f]*t[f],l+=s[f]*t[f];let h=Math.sqrt((o-i*i/n)*(a-r*r/n));if(h<1e-12)return 1;let c=(l-i*r/n)/h;return 1-Math.abs(c)}function Ra(s,t,n,e){let i=t.length;if(s.length<=2||e<=0)return s.map((o,a)=>a);let r=[0];for(let o=1;o<s.length;o++){let a=!0;for(let l of r)if(Pa(s[o],s[l],i,n)<e){a=!1;break}a&&r.push(o)}return r.length<2&&s.length>=2&&(r.includes(1)||r.push(1)),r}function Ma(s){return s.map(t=>Array.isArray(t)?{name:t[0],cls:t[1],params:t[2]||{}}:t)}async function za(s,t,n,e={}){let{ensemble:i=!0,ensembleSize:r=20,refit:o=!0,strategy:a="random",minDisagreement:l=.05,stacking:h="auto",metaEstimator:c=null,preprocess:f=!1,onProgress:u=null,...d}=e,m=Ma(s);if(m.length===0)throw new Cn("autoFit: at least one model is required");let p=null;if(f){let j=typeof f=="object"?f:{};p=new xa(j);let Y=Bn(t),T=On(n);t=p.fitTransform(Y,T)}let g={...d,onProgress:u},w;if(a==="portfolio")w=new qa(m,g);else if(a==="halving")w=new ka(m,g);else if(a==="progressive")w=new Ea(m,g);else if(a==="bayesian"){let j;try{j=Jn("@wlearn/bo").BayesianSearch}catch{throw new Cn('autoFit: strategy "bayesian" requires @wlearn/bo. Install: npm i @wlearn/bo')}w=new j(m,g)}else w=new Aa(m,g);let{leaderboard:b,bestResult:y}=await w.fit(t,n),_=b.ranked(),A=Bn(t),v=On(n),F=d.task||Ia(v),k=d.scoring||(F==="classification"?"accuracy":"r2"),O=d.cv||5,P=d.seed||42,it=null;if(i){u&&u({phase:"ensemble",message:"building ensemble"});let j=new Map,Y=new Map;for(let x of _)j.has(x.modelName)?Y.has(x.modelName)||Y.set(x.modelName,x):j.set(x.modelName,x);let T=[...j.values()],Z=new Set(T.map(x=>x.id));for(let x of Y.values()){if(T.length>=r*2)break;Z.has(x.id)||(T.push(x),Z.add(x.id))}for(let x of _){if(T.length>=r*2)break;Z.has(x.id)||(T.push(x),Z.add(x.id))}let ye=new Map;for(let x of m)ye.set(x.name,x.cls);let xe=T.map((x,ot)=>{let Wn=ye.get(x.modelName);return[`${x.modelName}_${ot}`,Wn,x.params]}),{oofPreds:Se}=await Sa(xe,A,v,{cv:O,seed:P,task:F}),_e=Ra(Se,v,F,l),Dn=_e.map(x=>Se[x]),Xn=_e.map(x=>xe[x]),{indices:Vn,weights:Un}=_a(Dn,v,{maxSize:r,scoring:k,task:F}),xt=Array.from(Vn,x=>Xn[x]),Yn=Un,Gn=new Set(xt.map(x=>x[0].split("_")[0]));if((h===!0||h==="auto"&&Gn.size>=3&&c)&&c){let x=Array.isArray(c)?c:["meta",c.cls||c,c.params||{}],ot=await Fa.create({estimators:xt,finalEstimator:x,passthrough:!0,task:F,cv:O,seed:P});await ot.fit(A,v),it=ot}else{let x=await va.create({estimators:xt,weights:Yn,voting:F==="classification"?"soft":void 0,task:F});await x.fit(A,v),it=x}}else o&&(it=await w.refitBest(t,n));return{model:it,preprocessor:p,leaderboard:_,bestParams:y.params,bestModelName:y.modelName,bestScore:y.meanScore}}jn.exports={autoFit:za}});var sc=S((Kc,$n)=>{var{sampleParam:Na,sampleConfig:La,randomConfigs:Ba,gridConfigs:Oa}=st(),{RandomSearch:Ca}=Jt(),{SuccessiveHalvingSearch:ja}=ee(),{PortfolioSearch:Ta,PortfolioStrategy:$a,getPortfolio:Da,PORTFOLIO:Xa}=ie(),{Leaderboard:Va}=Dt(),{autoFit:Ua}=Tn(),{Executor:Ya}=H(),{RandomStrategy:Ga}=Yt(),{HalvingStrategy:Wa}=Qt(),{ProgressiveStrategy:Ja}=pe(),{ProgressiveSearch:Ka}=we(),{detectTask:Ha,makeCandidateId:Za,seedFor:Qa,partialShuffle:tc,scorerGreaterIsBetter:ec}=M();$n.exports={sampleParam:Na,sampleConfig:La,randomConfigs:Ba,gridConfigs:Oa,RandomSearch:Ca,SuccessiveHalvingSearch:ja,PortfolioSearch:Ta,PortfolioStrategy:$a,getPortfolio:Da,PORTFOLIO:Xa,Leaderboard:Va,autoFit:Ua,Executor:Ya,RandomStrategy:Ga,HalvingStrategy:Wa,ProgressiveStrategy:Ja,ProgressiveSearch:Ka,detectTask:Ha,makeCandidateId:Za,seedFor:Qa,partialShuffle:tc,scorerGreaterIsBetter:ec}});return sc();})();
+var {sampleParam,sampleConfig,randomConfigs,gridConfigs,RandomSearch,SuccessiveHalvingSearch,PortfolioSearch,PortfolioStrategy,getPortfolio,PORTFOLIO,Leaderboard,autoFit,Executor,RandomStrategy,HalvingStrategy,ProgressiveStrategy,ProgressiveSearch,detectTask,makeCandidateId,seedFor,partialShuffle,scorerGreaterIsBetter}=__automl;export{sampleParam,sampleConfig,randomConfigs,gridConfigs,RandomSearch,SuccessiveHalvingSearch,PortfolioSearch,PortfolioStrategy,getPortfolio,PORTFOLIO,Leaderboard,autoFit,Executor,RandomStrategy,HalvingStrategy,ProgressiveStrategy,ProgressiveSearch,detectTask,makeCandidateId,seedFor,partialShuffle,scorerGreaterIsBetter};

package/package.json

@@ -1,27 +1,63 @@
 {
   "name": "@wlearn/automl",
-  "version": "0.1.0",
+  "version": "0.2.0",
   "description": "AutoML engine for wlearn: search space sampling, random search, successive halving, ensemble construction",
-  "type": "module",
+  "type": "commonjs",
   "main": "src/index.js",
   "exports": {
-    ".": "./src/index.js"
+    ".": {
+      "require": "./src/index.js",
+      "default": "./src/index.js"
+    }
   },
   "files": [
-    "src/"
+    "src/",
+    "dist/",
+    "LICENSE",
+    "NOTICE",
+    "README.md",
+    "CHANGELOG.md"
   ],
   "sideEffects": false,
   "dependencies": {
-    "@wlearn/core": "0.1.0",
-    "@wlearn/ensemble": "0.1.0",
-    "@wlearn/types": "0.1.0"
+    "@wlearn/core": "0.2.0",
+    "@wlearn/ensemble": "0.2.0"
   },
   "scripts": {
-    "test": "node --test test/*.js"
+    "test": "node --test test/*.js",
+    "build:browser": "bash scripts/build-browser.sh",
+    "prepack": "node scripts/prepack.js"
   },
   "publishConfig": {
     "access": "public"
   },
+  "keywords": [
+    "wlearn",
+    "automl",
+    "machine-learning",
+    "hyperparameter-optimization"
+  ],
   "author": "Anton Zemlyansky",
-  "license": "MIT"
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/wlearn-org/wlearn.git",
+    "directory": "packages/automl"
+  },
+  "license": "Apache-2.0",
+  "devDependencies": {
+    "esbuild": "^0.27.3",
+    "playwright": "^1.58.2"
+  },
+  "peerDependencies": {
+    "@wlearn/bo": "0.1.0"
+  },
+  "peerDependenciesMeta": {
+    "@wlearn/bo": {
+      "optional": true
+    }
+  },
+  "bugs": {
+    "url": "https://github.com/wlearn-org/wlearn/issues"
+  },
+  "homepage": "https://wlearn.org"
 }

package/src/auto-fit.js

@@ -1,10 +1,10 @@
-import { normalizeX, normalizeY, ValidationError, Preprocessor } from '@wlearn/core'
-import { getOofPredictions, caruanaSelect, VotingEnsemble, StackingEnsemble } from '@wlearn/ensemble'
-import { RandomSearch } from './search.js'
-import { SuccessiveHalvingSearch } from './halving.js'
-import { PortfolioSearch } from './portfolio.js'
-import { ProgressiveSearch } from './progressive.js'
-import { detectTask } from './common.js'
+const { normalizeX, normalizeY, ValidationError, Preprocessor } = require('@wlearn/core')
+const { getOofPredictions, caruanaSelect, VotingEnsemble, StackingEnsemble } = require('@wlearn/ensemble')
+const { RandomSearch } = require('./search.js')
+const { SuccessiveHalvingSearch } = require('./halving.js')
+const { PortfolioSearch } = require('./portfolio.js')
+const { ProgressiveSearch } = require('./progressive.js')
+const { detectTask } = require('./common.js')
 
 /**
  * Compute pairwise disagreement rate between two prediction vectors.
@@ -89,7 +89,7 @@ function _normalizeSpecs(models) {
  * @param {object} opts
  * @returns {Promise<{ model: object, leaderboard: object[], bestParams: object, bestModelName: string, bestScore: number }>}
  */
-export async function autoFit(models, X, y, opts = {}) {
+async function autoFit(models, X, y, opts = {}) {
   const {
     ensemble = true,
     ensembleSize = 20,
@@ -128,6 +128,16 @@ export async function autoFit(models, X, y, opts = {}) {
     search = new SuccessiveHalvingSearch(specs, searchOptsWithProgress)
   } else if (strategy === 'progressive') {
     search = new ProgressiveSearch(specs, searchOptsWithProgress)
+  } else if (strategy === 'bayesian') {
+    let BayesianSearch
+    try {
+      BayesianSearch = require('@wlearn/bo').BayesianSearch
+    } catch (e) {
+      throw new ValidationError(
+        'autoFit: strategy "bayesian" requires @wlearn/bo. Install: npm i @wlearn/bo'
+      )
+    }
+    search = new BayesianSearch(specs, searchOptsWithProgress)
   } else {
     search = new RandomSearch(specs, searchOptsWithProgress)
   }
@@ -259,3 +269,5 @@ export async function autoFit(models, X, y, opts = {}) {
   }
 }
 
+module.exports = { autoFit }
+

package/src/common.js

@@ -1,11 +1,11 @@
-import { makeLCG } from '@wlearn/core'
+const { makeLCG } = require('@wlearn/core')
 
 const { round } = Math
 
 /**
  * Detect task type from labels.
  */
-export function detectTask(y) {
+function detectTask(y) {
   if (y instanceof Int32Array) return 'classification'
   const unique = new Set()
   for (let i = 0; i < y.length; i++) {
@@ -18,7 +18,7 @@ export function detectTask(y) {
 /**
  * High-resolution timer.
  */
-export function now() {
+function now() {
   if (typeof performance !== 'undefined') return performance.now()
   return Date.now()
 }
@@ -46,7 +46,7 @@ function stableStringify(obj) {
 /**
  * Stable candidate ID from model label and params.
  */
-export function makeCandidateId(modelLabel, params) {
+function makeCandidateId(modelLabel, params) {
   return modelLabel + ':' + stableStringify(params)
 }
 
@@ -65,7 +65,7 @@ function hashString(str) {
 /**
  * Derive a deterministic seed from base seed, candidate ID, and fold index.
  */
-export function seedFor(candidateId, foldIdx, baseSeed) {
+function seedFor(candidateId, foldIdx, baseSeed) {
   const h = hashString(candidateId)
   // Mix: multiply-xor-shift
   let s = (baseSeed * 2654435761 + h * 40503 + foldIdx * 65537) & 0x7fffffff
@@ -77,7 +77,7 @@ export function seedFor(candidateId, foldIdx, baseSeed) {
  * Partial Fisher-Yates: shuffle only first k positions of indices array.
  * O(k) time, mutates indices in-place. Returns indices subarray [0..k-1].
  */
-export function partialShuffle(indices, k, rng) {
+function partialShuffle(indices, k, rng) {
   const n = indices.length
   const m = Math.min(k, n)
   for (let i = 0; i < m; i++) {
@@ -94,7 +94,7 @@ export function partialShuffle(indices, k, rng) {
  * All built-in scorers are greater-is-better (neg_mse, neg_mae are negated).
  * Custom functions default to true.
  */
-export function scorerGreaterIsBetter(scoring) {
+function scorerGreaterIsBetter(scoring) {
   if (typeof scoring === 'function') return true
   switch (scoring) {
     case 'accuracy':
@@ -106,3 +106,5 @@ export function scorerGreaterIsBetter(scoring) {
       return true
   }
 }
+
+module.exports = { detectTask, now, makeCandidateId, seedFor, partialShuffle, scorerGreaterIsBetter }

package/src/executor.js

@@ -1,6 +1,6 @@
-import { normalizeX, normalizeY, makeLCG, getScorer } from '@wlearn/core'
-import { Leaderboard } from './leaderboard.js'
-import { now, seedFor, partialShuffle } from './common.js'
+const { normalizeX, normalizeY, makeLCG, getScorer } = require('@wlearn/core')
+const { Leaderboard } = require('./leaderboard.js')
+const { now, seedFor, partialShuffle } = require('./common.js')
 
 const { ceil, min } = Math
 
@@ -35,7 +35,7 @@ function subsetY(y, indices) {
  * Evaluates candidates across all CV folds, applies budgets,
  * records results in a single Leaderboard instance.
  */
-export class Executor {
+class Executor {
   #folds
   #scorerFn
   #X
@@ -207,3 +207,5 @@ export class Executor {
     return { leaderboard: this.#leaderboard }
   }
 }
+
+module.exports = { Executor }

package/src/halving.js

@@ -1,15 +1,15 @@
-import { stratifiedKFold, kFold, normalizeX, normalizeY,
-  ValidationError } from '@wlearn/core'
-import { Executor } from './executor.js'
-import { HalvingStrategy } from './strategy-halving.js'
-import { detectTask, scorerGreaterIsBetter } from './common.js'
+const { stratifiedKFold, kFold, normalizeX, normalizeY,
+  ValidationError } = require('@wlearn/core')
+const { Executor } = require('./executor.js')
+const { HalvingStrategy } = require('./strategy-halving.js')
+const { detectTask, scorerGreaterIsBetter } = require('./common.js')
 
 /**
  * Successive halving search: multi-round elimination tournament.
  * Evaluates many candidates on small subsamples, progressively
  * eliminates the worst and increases resource allocation.
  */
-export class SuccessiveHalvingSearch {
+class SuccessiveHalvingSearch {
   #models
   #opts
   #leaderboard = null
@@ -93,3 +93,5 @@ export class SuccessiveHalvingSearch {
   get bestResult() { return this.#bestResult }
   get rounds() { return this.#rounds }
 }
+
+module.exports = { SuccessiveHalvingSearch }

package/src/index.js

@@ -1,12 +1,19 @@
-export { sampleParam, sampleConfig, randomConfigs, gridConfigs } from './sampler.js'
-export { RandomSearch } from './search.js'
-export { SuccessiveHalvingSearch } from './halving.js'
-export { PortfolioSearch, PortfolioStrategy, getPortfolio, PORTFOLIO } from './portfolio.js'
-export { Leaderboard } from './leaderboard.js'
-export { autoFit } from './auto-fit.js'
-export { Executor } from './executor.js'
-export { RandomStrategy } from './strategy-random.js'
-export { HalvingStrategy } from './strategy-halving.js'
-export { ProgressiveStrategy } from './strategy-progressive.js'
-export { ProgressiveSearch } from './progressive.js'
-export { detectTask, makeCandidateId, seedFor, partialShuffle, scorerGreaterIsBetter } from './common.js'
+const { sampleParam, sampleConfig, randomConfigs, gridConfigs } = require('./sampler.js')
+const { RandomSearch } = require('./search.js')
+const { SuccessiveHalvingSearch } = require('./halving.js')
+const { PortfolioSearch, PortfolioStrategy, getPortfolio, PORTFOLIO } = require('./portfolio.js')
+const { Leaderboard } = require('./leaderboard.js')
+const { autoFit } = require('./auto-fit.js')
+const { Executor } = require('./executor.js')
+const { RandomStrategy } = require('./strategy-random.js')
+const { HalvingStrategy } = require('./strategy-halving.js')
+const { ProgressiveStrategy } = require('./strategy-progressive.js')
+const { ProgressiveSearch } = require('./progressive.js')
+const { detectTask, makeCandidateId, seedFor, partialShuffle, scorerGreaterIsBetter } = require('./common.js')
+
+module.exports = {
+  sampleParam, sampleConfig, randomConfigs, gridConfigs,
+  RandomSearch, SuccessiveHalvingSearch, PortfolioSearch, PortfolioStrategy, getPortfolio, PORTFOLIO,
+  Leaderboard, autoFit, Executor, RandomStrategy, HalvingStrategy, ProgressiveStrategy, ProgressiveSearch,
+  detectTask, makeCandidateId, seedFor, partialShuffle, scorerGreaterIsBetter,
+}

package/src/leaderboard.js

@@ -1,7 +1,7 @@
 /**
  * Tracks and ranks candidate evaluation results.
  */
-export class Leaderboard {
+class Leaderboard {
   #entries = []
   #nextId = 0
   #dirty = true
@@ -104,3 +104,5 @@ export class Leaderboard {
     return this.#entries.length
   }
 }
+
+module.exports = { Leaderboard }