xy-scale 1.4.44 → 1.4.46

package/README.md

@@ -29,7 +29,8 @@ Builds supervised-learning datasets and splits them into training and testing ar
 #### Parameters
 
 - `arrObj` (Array<Object>): Source dataset.
-- `trainingSplit` (Number, optional): Fraction of rows used for training. Default: `0.8`.
+- `trainSize` (Number, required).
+- `testSize` (Number, required).
 - `yCallbackFunc` (Function, optional): Builds the output object for each row. Returning `null` or `undefined` skips the row.
 - `xCallbackFunc` (Function, optional): Builds the feature object for each row. Returning `null` or `undefined` skips the row.
 - `validateRows` (Function, optional): Extra row filter executed before the callbacks.

package/dist/xy-scale.min.js

@@ -1 +1 @@
-var XY_Scale;(()=>{"use strict";var e={d:(r,t)=>{for(var n in t)e.o(t,n)&&!e.o(r,n)&&Object.defineProperty(r,n,{enumerable:!0,get:t[n]})},o:(e,r)=>Object.prototype.hasOwnProperty.call(e,r),r:e=>{"undefined"!=typeof Symbol&&Symbol.toStringTag&&Object.defineProperty(e,Symbol.toStringTag,{value:"Module"}),Object.defineProperty(e,"__esModule",{value:!0})}},r={};e.r(r),e.d(r,{arrayShuffle:()=>t,arrayShuffleXY:()=>n,arrayToTimesteps:()=>b,parseProductionX:()=>g,parseTrainingXY:()=>w});const t=e=>{const r=[...e];for(let e=r.length-1;e>0;e--){const t=Math.floor(Math.random()*(e+1));[r[e],r[t]]=[r[t],r[e]]}return r},n=(e,r)=>{if(!Array.isArray(e)||!Array.isArray(r))throw new TypeError("Both X and Y must be arrays");if(e.length!==r.length)throw new Error("X and Y must have the same length");const t=[...e],n=[...r];for(let e=t.length-1;e>0;e--){const r=Math.floor(Math.random()*(e+1));[t[e],t[r]]=[t[r],t[e]],[n[e],n[r]]=[n[r],n[e]]}return{X:t,Y:n}},o=e=>"string"==typeof e||"boolean"==typeof e||("number"==typeof e?Number.isFinite(e):!!Array.isArray(e)&&e.every(o)),a=({random:e,seed:r}={})=>"function"==typeof e?e:Number.isInteger(r)?(e=>{let r=e>>>0;return()=>{r+=1831565813;let e=Math.imul(r^r>>>15,1|r);return e^=e+Math.imul(e^e>>>7,61|e),((e^e>>>14)>>>0)/4294967296}})(r):Math.random,l=(e,r=Math.random)=>{for(let t=e.length-1;t>0;t--){const n=Math.floor(r()*(t+1));[e[t],e[n]]=[e[n],e[t]]}return e},i=(e,r,t=Math.random)=>{if(r>e.length)throw new Error("Cannot sample more items than available without replacement.");const n=[...e];return l(n,t),n.slice(0,r)},s=(e,r,t=Math.random)=>{if(0===e.length)throw new Error("Cannot sample from an empty array.");const n=[];for(let o=0;o<r;o++){const r=Math.floor(t()*e.length);n.push(e[r])}return n},u=(e,r)=>r?structuredClone(e):e,c=(e,r,{cloneX:t=!1}={})=>{((e,r)=>{if(!Array.isArray(e)||!Array.isArray(r))throw new Error("X and Y must be arrays.");if(e.length!==r.length)throw new Error("X and Y must have the same length.");if(0===e.length)throw new Error("X and Y cannot be empty.")})(e,r);const n=new Map;return r.forEach(((r,a)=>{const l=(e=>{if(!o(e))throw new Error("Invalid Y label. Allowed types: finite numbers, strings, booleans, or nested arrays of those.");return JSON.stringify(e)})(r);n.has(l)||n.set(l,[]),n.get(l).push({x:u(e[a],t),y:r})})),n},f=e=>null!=e&&Number.isFinite(e),h=e=>null!==e&&"object"==typeof e&&!Array.isArray(e)&&Object.keys(e).length>0,d=(e,{min:r=-1/0,max:t=1/0},n)=>{if(!Array.isArray(e))throw new Error(`Invalid property. "${n}" expected an array.`);if(e.length<r)throw new Error(`Invalid property value. Array "${n}" expected at least ${r} items.`);if(e.length>t)throw new Error(`Invalid property value. Array "${n}" expected at max ${t} items.`);return!0},y=e=>{if(!h(e))throw new Error("The first item in arrObj is expeted to be a key par object.");const r={},t={};for(const[n,o]of Object.entries(e))"number"==typeof o?r[n]=o:t[n]=o;if(Object.keys(r).length>0&&m(r,"validateFirstRow"))throw new Error("Invalid numeric value at index 0.");if(Object.keys(t).length>0&&p(t,"validateFirstRow"))throw new Error("Invalid non-numeric value at index 0.");return!0},m=(e,r)=>{if(null==r)throw new Error('[hasInvalidNumbers] Missing required param "callerName".');if(!h(e))throw new Error(`[${r}:hasInvalidNumbers] Received an invalid "list" param — expected a non-empty key-pair object, got: ${JSON.stringify(e)}.`);for(const[t,n]of Object.entries(e))if(!f(n))return console.error(`[${r}:hasInvalidNumbers] property "${t}" only accept numbers. Invalid value is "${n}" and invalid type is "${typeof n}".`),!0;return!1},p=(e,r)=>{if(null==r)throw new Error('[hasNullOrUndefined] Missing required param "callerName".');if(!h(e))throw new Error(`[${r}:hasNullOrUndefined] Received an invalid "list" param — expected a non-empty key-pair object, got: ${JSON.stringify(e)}.`);for(const[t,n]of Object.entries(e))if(h(n)){for(const[e,o]of Object.entries(n))if(null==o)return console.error(`[${r}:hasNullOrUndefined] Null or undefined value detected for key "${t}.${e}".\n${String(o)}`),!0}else if(null==n)return console.error(`[${r}:hasNullOrUndefined] Null or undefined value detected for key "${t}".\n${String(n)}`),!0;return!1},w=({arrObj:e=[],trainingSplit:r=.8,yCallbackFunc:n=e=>e,xCallbackFunc:o=e=>e,validateRows:f=()=>!0,shuffle:h=!1,balancing:p="",state:w={}})=>{d(e,{min:2},"parseTrainingXY"),y(e[0]);let g=[],b=[],v=null,x=null;const X={};for(let r=0;r<e.length;r++)try{if(!f({objRow:e,index:r,state:w}))continue;const t=o({objRow:e,index:r,state:w}),a=n({objRow:e,index:r,state:w});if(null==t||null==a)continue;if(m(t,"parseTrainingXY"))throw new Error('Invalid numeric value returned from "xCallbackFunc".');if(null===v&&(v=Object.keys(t)),null===x){x=Object.keys(a);for(let e=0;e<x.length;e++)X[x[e]]={}}const l=v.length,i=x.length,s=new Array(l),u=new Array(i);for(let e=0;e<l;e++){const r=v[e];s[e]=t[r]}for(let e=0;e<i;e++){const r=x[e],t=a[r];u[e]=t;const n=Array.isArray(t)?JSON.stringify(t):String(t);X[r][n]=(X[r][n]??0)+1}g.push(s),b.push(u)}catch(e){throw new Error(`[BUG] - Skipped row index=${r}: ${e.message}`)}if(h){const e=new Array(g.length);for(let r=0;r<g.length;r++)e[r]={x:g[r],y:b[r]};const r=t(e);g=new Array(r.length),b=new Array(r.length);for(let e=0;e<r.length;e++)g[e]=r[e].x,b[e]=r[e].y}const E={keyNames:v??[]},A={keyNames:x??[],labelCounts:X},j=Math.floor(g.length*r);let O=g.slice(0,j),$=b.slice(0,j),k=g.slice(j),N=b.slice(j);if(p){let e;if("oversample"===p)e=((e,r,t={})=>{const{random:n,seed:o,shuffleResult:i=!0,cloneX:f=!1}=t,h=a({random:n,seed:o}),d=c(e,r,{cloneX:f}),y=[...d.values()].map((e=>e.length)),m=Math.max(...y),p=[];for(const e of d.values()){const r=[...e],t=m-r.length,n=t>0?s(e,t,h).map((e=>({x:u(e.x,f),y:e.y}))):[];p.push(...r,...n)}return i&&l(p,h),{X:p.map((({x:e})=>e)),Y:p.map((({y:e})=>e))}})(O,$),O=e.X,$=e.Y;else{if("undersample"!==p)throw Error('balancing argument only accepts "", "oversample" and "undersample". Defaults to "".');e=((e,r,t={})=>{const{random:n,seed:o,shuffleResult:s=!0,cloneX:f=!1}=t,h=a({random:n,seed:o}),d=c(e,r,{cloneX:f}),y=[...d.values()].map((e=>e.length)),m=Math.min(...y),p=[];for(const e of d.values()){const r=i(e,m,h).map((e=>({x:u(e.x,f),y:e.y})));p.push(...r)}return s&&l(p,h),{X:p.map((({x:e})=>e)),Y:p.map((({y:e})=>e))}})(O,$),O=e.X,$=e.Y}}return{trainX:O,trainY:$,testX:k,testY:N,configX:E,configY:A}},g=({arrObj:e=[],xCallbackFunc:r=e=>e,yCallbackFunc:n=null,validateRows:o=()=>!0,shuffle:a=!1,state:l={}})=>{let i=[],s=null;if(d(e,{min:1},"parseProductionX"),y(e[0]),null!=n)throw new Error('The property "yCallbackFunc" must not be set in "parseProductionX".');for(let t=0;t<e.length;t++)try{if(!o({objRow:e,index:t,state:l}))continue;const n=r({objRow:e,index:t,state:l});if(null==n)continue;if(m(n,"parseProductionX"))throw new Error('Invalid numeric value returned from "xCallbackFunc".');null===s&&(s=Object.keys(n));const a=s.length,u=new Array(a);for(let e=0;e<a;e++){const r=s[e];u[e]=n[r]}i.push(u)}catch(e){throw new Error(`[BUG] - Skipped row index=${t}: ${e.message}`)}return a&&(i=t(i)),{X:i,configX:{keyNames:s??[]}}},b=(e,r,t=1)=>{if(!Array.isArray(e))throw new Error("arr must be an array");if(!Number.isInteger(r)||r<=0)throw new Error("timeSteps must be a positive integer");if(!Number.isInteger(t)||t<=0)throw new Error("step must be a positive integer");if(r>e.length)return[];const n=[];for(let o=0;o<=e.length-r;o+=t)n.push(e.slice(o,o+r));return n};XY_Scale=r})();
+var XY_Scale;(()=>{"use strict";var e={d:(r,t)=>{for(var n in t)e.o(t,n)&&!e.o(r,n)&&Object.defineProperty(r,n,{enumerable:!0,get:t[n]})},o:(e,r)=>Object.prototype.hasOwnProperty.call(e,r),r:e=>{"undefined"!=typeof Symbol&&Symbol.toStringTag&&Object.defineProperty(e,Symbol.toStringTag,{value:"Module"}),Object.defineProperty(e,"__esModule",{value:!0})}},r={};e.r(r),e.d(r,{arrayShuffle:()=>t,arrayToTimesteps:()=>h,parseProductionX:()=>w,parseTrainingXY:()=>d,zscore:()=>f});const t=e=>{const r=[...e];for(let e=r.length-1;e>0;e--){const t=Math.floor(Math.random()*(e+1));[r[e],r[t]]=[r[t],r[e]]}return r},n=e=>null!=e&&Number.isFinite(e),o=e=>Number.isInteger(e)&&e>0,a=e=>null!==e&&"object"==typeof e&&!Array.isArray(e)&&Object.keys(e).length>0,i=(e,{min:r=-1/0,max:t=1/0},n)=>{if(!Array.isArray(e))throw new Error(`Invalid property. "${n}" expected an array.`);if(e.length<r)throw new Error(`Invalid property value. Array "${n}" expected at least ${r} items.`);if(e.length>t)throw new Error(`Invalid property value. Array "${n}" expected at max ${t} items.`);return!0},l=e=>{if(!a(e))throw new Error("The first item in arrObj is expeted to be a key par object.");const r={},t={};for(const[n,o]of Object.entries(e))"number"==typeof o?r[n]=o:t[n]=o;if(Object.keys(r).length>0&&s(r,"validateFirstRow"))throw new Error("Invalid numeric value at index 0.");if(Object.keys(t).length>0&&c(t,"validateFirstRow"))throw new Error("Invalid non-numeric value at index 0.");return!0},s=(e,r)=>{if(null==r)throw new Error('[hasInvalidNumbers] Missing required param "callerName".');if(!a(e))throw new Error(`[${r}:hasInvalidNumbers] Received an invalid "list" param — expected a non-empty key-pair object, got: ${JSON.stringify(e)}.`);for(const[t,o]of Object.entries(e))if(!n(o))return console.error(`[${r}:hasInvalidNumbers] property "${t}" only accept numbers. Invalid value is "${o}" and invalid type is "${typeof o}".`),!0;return!1},c=(e,r)=>{if(null==r)throw new Error('[hasNullOrUndefined] Missing required param "callerName".');if(!a(e))throw new Error(`[${r}:hasNullOrUndefined] Received an invalid "list" param — expected a non-empty key-pair object, got: ${JSON.stringify(e)}.`);for(const[t,n]of Object.entries(e))if(a(n)){for(const[e,o]of Object.entries(n))if(null==o)return console.error(`[${r}:hasNullOrUndefined] Null or undefined value detected for key "${t}.${e}".\n${String(o)}`),!0}else if(null==n)return console.error(`[${r}:hasNullOrUndefined] Null or undefined value detected for key "${t}".\n${String(n)}`),!0;return!1},u=e=>Array.from(e),y=(e,r)=>{const t=new Float64Array(r),n=new Float64Array(r),o=new Float64Array(r),a=new Uint32Array(r),i=Array.isArray(e?.mean)?e.mean:[],l=Array.isArray(e?.std)?e.std:[],s=Array.isArray(e?.scale)?e.scale:[],c=Array.isArray(e?.count)?e.count:[];for(let e=0;e<r;e++){const r=Number(i[e]),u=Number(l[e]),y=Number(s[e]),f=Number(c[e]);t[e]=Number.isFinite(r)?r:0,n[e]=Number.isFinite(u)&&u>0?u:0,a[e]=Number.isFinite(f)&&f>0?f:0,Number.isFinite(y)&&y>0?(o[e]=y,n[e]>0||(n[e]=1/y)):n[e]>0?o[e]=1/n[e]:o[e]=0}return{mean:t,std:n,scale:o,count:a}},f=(e,r=null)=>{if(!Array.isArray(e))throw new TypeError('[zscore] "arr" must be a 2D array.');if(0===e.length){const e=y(r??{},0);return{stats:{mean:u(e.mean),std:u(e.std),scale:u(e.scale),count:u(e.count)},data:[]}}const t=e[0];if(!Array.isArray(t))throw new TypeError('[zscore] "arr" must be a 2D array of rows.');const n=t.length,o=r?y(r,n):((e,r)=>{const t=e.length,n=new Float64Array(r),o=new Float64Array(r),a=new Uint32Array(r);for(let i=0;i<t;i++){const t=e[i];if(!Array.isArray(t))throw new TypeError(`[zscore] Invalid row at index=${i}. Expected an array.`);if(t.length!==r)throw new Error(`[zscore] Inconsistent row size at index=${i}. Expected ${r}, got ${t.length}.`);for(let e=0;e<r;e++){const r=Number(t[e]);if(!Number.isFinite(r))continue;a[e]++;const i=r-n[e];n[e]+=i/a[e],o[e]+=i*(r-n[e])}}const i=new Float64Array(r),l=new Float64Array(r);for(let e=0;e<r;e++){const r=a[e]>1?o[e]/a[e]:0,t=r>0?Math.sqrt(r):0;i[e]=t,l[e]=t>0?1/t:0}return{mean:n,std:i,scale:l,count:a}})(e,n),a=((e,r)=>{const t=e.length,n=r.mean.length,o=new Array(t);for(let a=0;a<t;a++){const t=e[a];if(!Array.isArray(t))throw new TypeError(`[zscore] Invalid row at index=${a}. Expected an array.`);if(t.length!==n)throw new Error(`[zscore] Inconsistent row size at index=${a}. Expected ${n}, got ${t.length}.`);const i=new Array(n);for(let e=0;e<n;e++){const n=Number(t[e]),o=r.scale[e];i[e]=Number.isFinite(n)&&o>0?(n-r.mean[e])*o:0}o[a]=i}return o})(e,o);return{stats:{mean:u(o.mean),std:u(o.std),scale:u(o.scale),count:u(o.count)},data:a}},d=({arrObj:e=[],trainSize:r=null,testSize:n=null,yCallbackFunc:a=e=>e,xCallbackFunc:c=e=>e,validateRows:u=()=>!0,shuffle:y=!1,state:d={},showSource:w=!1,scaling:h=null})=>{i(e,{min:2},"parseTrainingXY"),l(e[0]);const p=e.length;if((({arrObjSize:e,trainSize:r,testSize:t})=>{if(!o(r))throw new Error(`Invalid property: "trainSize" (${r}) must be a non-negative integer.`);if(!o(t))throw new Error(`Invalid property: "testSize" (${t}) must be a non-negative integer.`);if(!Number.isInteger(e)||e<0)throw new Error(`Invalid property: "arrObjSize" (${e}) must be a non-negative integer.`);if(e<r+t)throw new Error(`Invalid property: The sum of "trainSize" + "testSize" (${r+t}) must not be larger than "arrObj.length" (${e}).`)})({arrObjSize:p,trainSize:r,testSize:n}),![null,"zscore"].includes(h))throw new Error('Invalid "scaling" property. Accepting null or "zscore".');const g=r+n;let m=[],b=[],A=[],v=null,$=null;const E={};for(let r=0;r<e.length;r++)try{if(!u({objRow:e,index:r,state:d}))continue;const t=c({objRow:e,index:r,state:d}),n=a({objRow:e,index:r,state:d});if(null==t||null==n)continue;if(s(t,"parseTrainingXY"))throw new Error('Invalid numeric value returned from "xCallbackFunc".');if(null===v&&(v=Object.keys(t)),null===$){$=Object.keys(n);for(let e=0;e<$.length;e++)E[$[e]]={}}const o=v.length,i=$.length,l=new Array(o),y=new Array(i);for(let e=0;e<o;e++){const r=v[e];l[e]=t[r]}for(let e=0;e<i;e++){const r=$[e],t=n[r];y[e]=t;const o=Array.isArray(t)?JSON.stringify(t):String(t);E[r][o]=(E[r][o]??0)+1}m.push(l),b.push(y),w&&A.push(e[r])}catch(e){throw new Error(`[BUG] - Skipped row index=${r}: ${e.message}`)}if(y){const e=new Array(m.length);for(let r=0;r<m.length;r++)e[r]={x:m[r],y:b[r]},w&&(e[r].source=A[r]);const r=t(e);m=new Array(r.length),b=new Array(r.length),w&&(A=new Array(r.length));for(let e=0;e<r.length;e++)m[e]=r[e].x,b[e]=r[e].y,w&&(A[e]=r[e].source)}const N={keyNames:v??[]},S={keyNames:$??[],labelCounts:E},x=p-g;m.splice(0,x),b.splice(0,x);let j=m.slice(0,r),z=b.slice(0,r),O=null;if("zscore"===h){let e=f(j);O=e.stats,j=e.data,e=null}let I=m.slice(-n);"zscore"===h&&(I=f(I,O).data);let k,F,T=b.slice(-n);return m=null,b=null,w&&(A.splice(0,x),k=A.slice(0,r),F=A.slice(-n),A=null),{trainX:j,trainY:z,testX:I,testY:T,configX:N,configY:S,trainSource:k,testSource:F,stats:O}},w=({arrObj:e=[],xCallbackFunc:r=e=>e,yCallbackFunc:n=null,validateRows:o=()=>!0,shuffle:a=!1,state:c={},showSource:u=!1,scaling:y=null,stats:d})=>{let w=[],h=[],p=null;if(i(e,{min:1},"parseProductionX"),l(e[0]),![null,"zscore"].includes(y))throw new Error('Invalid "scaling" property. Accepting null or "zscore".');if(e.length,null!=n)throw new Error('The property "yCallbackFunc" must not be set in "parseProductionX".');for(let t=0;t<e.length;t++)try{if(!o({objRow:e,index:t,state:c}))continue;const n=r({objRow:e,index:t,state:c});if(null==n)continue;if(s(n,"parseProductionX"))throw new Error('Invalid numeric value returned from "xCallbackFunc".');null===p&&(p=Object.keys(n));const a=p.length,i=new Array(a);for(let e=0;e<a;e++){const r=p[e];i[e]=n[r]}w.push(i),u&&h.push(e[t])}catch(e){throw new Error(`[BUG] - Skipped row index=${t}: ${e.message}`)}if(a){const e=new Array(w.length);for(let r=0;r<w.length;r++)e[r]={x:w[r]},u&&(e[r].source=h[r]);const r=t(e);w=new Array(r.length),u&&(h=new Array(r.length));for(let e=0;e<r.length;e++)w[e]=r[e].x,u&&(h[e]=r[e].source)}const g={keyNames:p??[]};return"zscore"===y&&(w=f(w,d).data),{X:w,source:h,configX:g}},h=(e,r,t=1)=>{if(!Array.isArray(e))throw new Error("arr must be an array");if(!Number.isInteger(r)||r<=0)throw new Error("timeSteps must be a positive integer");if(!Number.isInteger(t)||t<=0)throw new Error("step must be a positive integer");if(r>e.length)return[];const n=[];for(let o=0;o<=e.length-r;o+=t)n.push(e.slice(o,o+r));return n};XY_Scale=r})();

package/index.js

@@ -1,5 +1,6 @@
 import { parseTrainingXY, parseProductionX } from "./src/datasets.js"
 import {arrayToTimesteps } from "./src/timeSteps.js"
 import { arrayShuffle } from "./src/utilities.js"
+import { zscore } from "./src/zscore.js"
 
-export { parseTrainingXY, parseProductionX, arrayToTimesteps, arrayShuffle }
+export { parseTrainingXY, parseProductionX, arrayToTimesteps, arrayShuffle, zscore }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "xy-scale",
-  "version": "1.4.44",
+  "version": "1.4.46",
   "main": "./index.js",
   "type": "module",
   "scripts": {

package/src/datasets.js

@@ -1,23 +1,35 @@
 import { arrayShuffle } from "./utilities.js";
-import { validateFirstRow, validateArray, hasInvalidNumbers } from "./validators.js";
+import { validateFirstRow, validateArray, hasInvalidNumbers, validateSizes } from "./validators.js";
+import {zscore} from './zscore.js'
 
 export const parseTrainingXY = ({
     arrObj = [],
-    trainingSplit = 0.8,
+    trainSize = null,
+    testSize = null,
     yCallbackFunc = row => row,
     xCallbackFunc = row => row,
     validateRows = () => true,
     shuffle = false,
     state = {},
+    showSource = false,
+    scaling = null
 }) => {
     validateArray(arrObj, { min: 2 }, 'parseTrainingXY');
     validateFirstRow(arrObj[0]);
 
+    const arrObjSize = arrObj.length;
+    
+    validateSizes({arrObjSize, trainSize, testSize});
+
+    if(![null, 'zscore'].includes(scaling)) {
+        throw new Error(`Invalid "scaling" property. Accepting null or "zscore".`)
+    }
+
+    const totalSize = trainSize + testSize;
     let flatX = [];
     let flatY = [];
     let source = [];
 
-
     let keyNamesX = null;
     let keyNamesY = null;
 
@@ -75,7 +87,11 @@ export const parseTrainingXY = ({
 
             flatX.push(rowX);
             flatY.push(rowY);
-            source.push(arrObj[x])
+
+            if(showSource) {
+                source.push(arrObj[x])
+            }
+            
 
         } catch(err) {
             throw new Error(`[BUG] - Skipped row index=${x}: ${err.message}`);
@@ -88,21 +104,31 @@ export const parseTrainingXY = ({
         for (let i = 0; i < flatX.length; i++) {
             merged[i] = {
                 x: flatX[i],
-                y: flatY[i],
-                source: source[i]
+                y: flatY[i]
             };
+
+            if(showSource) {
+                merged[i].source = source[i]
+            }
         }
 
         const shuffled = arrayShuffle(merged);
 
         flatX = new Array(shuffled.length);
         flatY = new Array(shuffled.length);
-        source = new Array(shuffled.length)
+
+        if(showSource) {
+            source = new Array(shuffled.length)
+        }
+        
 
         for (let i = 0; i < shuffled.length; i++) {
             flatX[i] = shuffled[i].x;
             flatY[i] = shuffled[i].y;
-            source[i] = shuffled[i].source
+
+            if(showSource) {
+                source[i] = shuffled[i].source;
+            }
         }
     }
 
@@ -115,14 +141,44 @@ export const parseTrainingXY = ({
         labelCounts,
     };
 
-    const splitIndex = Math.floor(flatX.length * trainingSplit);
+    const startSize = arrObjSize - totalSize
+
+    flatX.splice(0, startSize) //keeps the last items
+    flatY.splice(0, startSize) //keeps the last items
+
+    let trainX = flatX.slice(0, trainSize);
+    let trainY = flatY.slice(0, trainSize);
+    
+    let stats = null
+
+    if(scaling === 'zscore') {
+        let trainNormalized = zscore(trainX)
+        stats = trainNormalized.stats
+        trainX = trainNormalized.data
+        trainNormalized = null
+    }
+
+    let testX = flatX.slice(-testSize);
+
+    if(scaling === 'zscore') {
+        testX = zscore(testX, stats).data
+    }
+
+    let testY = flatY.slice(-testSize);
 
-    let trainX = flatX.slice(0, splitIndex);
-    let trainY = flatY.slice(0, splitIndex);
-    let testX = flatX.slice(splitIndex);
-    let testY = flatY.slice(splitIndex);
-    let trainSource = source.slice(0, splitIndex);
-    let testSource = source.slice(splitIndex);
+
+    flatX = null
+    flatY = null
+
+    let trainSource
+    let testSource
+
+    if(showSource) {
+        source.splice(0, startSize) //keeps the last items
+        trainSource = source.slice(0, trainSize);
+        testSource = source.slice(-testSize);
+        source = null
+    }
 
     return {
         trainX,
@@ -132,8 +188,9 @@ export const parseTrainingXY = ({
         configX,
         configY,
         trainSource,
-        testSource
-    };
+        testSource,
+        stats
+    }
 };
 export const parseProductionX = ({
     arrObj = [],
@@ -142,6 +199,9 @@ export const parseProductionX = ({
     validateRows = () => true,
     shuffle = false,
     state = {},
+    showSource = false,
+    scaling = null,
+    stats
 }) => {
     let flatX = [];
     let source = [];
@@ -150,6 +210,12 @@ export const parseProductionX = ({
     validateArray(arrObj, { min: 1 }, 'parseProductionX');
     validateFirstRow(arrObj[0]);
 
+    if(![null, 'zscore'].includes(scaling)) {
+        throw new Error(`Invalid "scaling" property. Accepting null or "zscore".`)
+    }
+
+    const arrObjSize = arrObj.length
+
     if (yCallbackFunc != null) {
         throw new Error('The property "yCallbackFunc" must not be set in "parseProductionX".');
     }
@@ -179,7 +245,11 @@ export const parseProductionX = ({
             }
 
             flatX.push(rowX);
-            source.push(arrObj[x])
+
+            if(showSource) {
+                source.push(arrObj[x])
+            }
+            
 
         } catch(err) {
             throw new Error(`[BUG] - Skipped row index=${x}: ${err.message}`);
@@ -191,19 +261,29 @@ export const parseProductionX = ({
 
         for (let i = 0; i < flatX.length; i++) {
             merged[i] = {
-                x: flatX[i],
-                source: source[i]
+                x: flatX[i]
             };
+
+            if(showSource) {
+                merged[i].source = source[i]
+            }
         }
 
         const shuffled = arrayShuffle(merged);
 
         flatX = new Array(shuffled.length);
-        source = new Array(shuffled.length)
+
+        if(showSource) {
+            source = new Array(shuffled.length)
+        }
+        
 
         for (let i = 0; i < shuffled.length; i++) {
             flatX[i] = shuffled[i].x;
-            source[i] = shuffled[i].source
+
+            if(showSource) {
+                source[i] = shuffled[i].source
+            }
         }
     }
 
@@ -211,6 +291,10 @@ export const parseProductionX = ({
         keyNames: keyNamesX ?? [],
     };
 
+    if(scaling === 'zscore') {
+        flatX = zscore(flatX, stats).data
+    }
+
     return {
         X: flatX,
         source,

package/src/validators.js

@@ -1,6 +1,8 @@
 
 export const isNumber = v => v != null && Number.isFinite(v)
 
+export const isPositiveInteger = value =>  Number.isInteger(value) && value > 0
+
 export const isKeyPairObject = param => {
   return (
     param !== null &&
@@ -132,4 +134,24 @@ export const arraysAreNotEqualSize = (list, callerName) => {
     }
 
     return false
+}
+
+export const validateSizes = ({arrObjSize, trainSize, testSize}) => {
+
+    if(!isPositiveInteger(trainSize)) {
+        throw new Error(`Invalid property: "trainSize" (${trainSize}) must be a non-negative integer.`)
+    }
+    if(!isPositiveInteger(testSize)) {
+         throw new Error(`Invalid property: "testSize" (${testSize}) must be a non-negative integer.`)
+    }
+
+    if (!Number.isInteger(arrObjSize) || arrObjSize < 0) {
+        throw new Error(`Invalid property: "arrObjSize" (${arrObjSize}) must be a non-negative integer.`)
+    }
+
+    if(arrObjSize < (trainSize + testSize)) {
+        throw new Error(`Invalid property: The sum of "trainSize" + "testSize" (${trainSize + testSize}) must not be larger than "arrObj.length" (${arrObjSize}).`)
+    }
+
+    return true
 }

package/src/zscore.js

@@ -0,0 +1,145 @@
+const toArray = typedOrArray => Array.from(typedOrArray);
+
+const normalizeStats = (stats, cols) => {
+  const mean = new Float64Array(cols);
+  const std = new Float64Array(cols);
+  const scale = new Float64Array(cols);
+  const count = new Uint32Array(cols);
+
+  const meanIn = Array.isArray(stats?.mean) ? stats.mean : [];
+  const stdIn = Array.isArray(stats?.std) ? stats.std : [];
+  const scaleIn = Array.isArray(stats?.scale) ? stats.scale : [];
+  const countIn = Array.isArray(stats?.count) ? stats.count : [];
+
+  for (let j = 0; j < cols; j++) {
+    const m = Number(meanIn[j]);
+    const s = Number(stdIn[j]);
+    const sc = Number(scaleIn[j]);
+    const c = Number(countIn[j]);
+
+    mean[j] = Number.isFinite(m) ? m : 0;
+    std[j] = Number.isFinite(s) && s > 0 ? s : 0;
+    count[j] = Number.isFinite(c) && c > 0 ? c : 0;
+
+    if (Number.isFinite(sc) && sc > 0) {
+      scale[j] = sc;
+      if (!(std[j] > 0)) { std[j] = 1 / sc; }
+    } else if (std[j] > 0) {
+      scale[j] = 1 / std[j];
+    } else {
+      scale[j] = 0;
+    }
+  }
+
+  return { mean, std, scale, count };
+};
+
+const fitStats = (arr, cols) => {
+  const rows = arr.length;
+  const mean = new Float64Array(cols);
+  const m2 = new Float64Array(cols);
+  const count = new Uint32Array(cols);
+
+  for (let i = 0; i < rows; i++) {
+    const row = arr[i];
+
+    if (!Array.isArray(row)) {
+      throw new TypeError(`[zscore] Invalid row at index=${i}. Expected an array.`);
+    }
+    if (row.length !== cols) {
+      throw new Error(`[zscore] Inconsistent row size at index=${i}. Expected ${cols}, got ${row.length}.`);
+    }
+
+    for (let j = 0; j < cols; j++) {
+      const x = Number(row[j]);
+      if (!Number.isFinite(x)) { continue; }
+
+      count[j]++;
+
+      const delta = x - mean[j];
+      mean[j] += delta / count[j];
+      m2[j] += delta * (x - mean[j]);
+    }
+  }
+
+  const std = new Float64Array(cols);
+  const scale = new Float64Array(cols);
+
+  for (let j = 0; j < cols; j++) {
+    const variance = count[j] > 1 ? (m2[j] / count[j]) : 0;
+    const s = variance > 0 ? Math.sqrt(variance) : 0;
+    std[j] = s;
+    scale[j] = s > 0 ? (1 / s) : 0;
+  }
+
+  return { mean, std, scale, count };
+};
+
+const scaleFromStats = (arr, normalizedStats) => {
+  const rows = arr.length;
+  const cols = normalizedStats.mean.length;
+  const out = new Array(rows);
+
+  for (let i = 0; i < rows; i++) {
+    const row = arr[i];
+
+    if (!Array.isArray(row)) {
+      throw new TypeError(`[zscore] Invalid row at index=${i}. Expected an array.`);
+    }
+    if (row.length !== cols) {
+      throw new Error(`[zscore] Inconsistent row size at index=${i}. Expected ${cols}, got ${row.length}.`);
+    }
+
+    const normalized = new Array(cols);
+    for (let j = 0; j < cols; j++) {
+      const x = Number(row[j]);
+      const sc = normalizedStats.scale[j];
+
+      normalized[j] = (Number.isFinite(x) && sc > 0)
+        ? (x - normalizedStats.mean[j]) * sc
+        : 0;
+    }
+
+    out[i] = normalized;
+  }
+
+  return out;
+};
+
+export const zscore = (arr, stats = null) => {
+  if (!Array.isArray(arr)) {
+    throw new TypeError('[zscore] "arr" must be a 2D array.');
+  }
+
+  if (arr.length === 0) {
+    const normalizedStats = normalizeStats(stats ?? {}, 0);
+    return {
+      stats: {
+        mean: toArray(normalizedStats.mean),
+        std: toArray(normalizedStats.std),
+        scale: toArray(normalizedStats.scale),
+        count: toArray(normalizedStats.count)
+      },
+      data: []
+    };
+  }
+
+  const firstRow = arr[0];
+  if (!Array.isArray(firstRow)) {
+    throw new TypeError('[zscore] "arr" must be a 2D array of rows.');
+  }
+
+  const cols = firstRow.length;
+  const normalizedStats = stats ? normalizeStats(stats, cols) : fitStats(arr, cols);
+  const data = scaleFromStats(arr, normalizedStats);
+
+  return {
+    stats: {
+      mean: toArray(normalizedStats.mean),
+      std: toArray(normalizedStats.std),
+      scale: toArray(normalizedStats.scale),
+      count: toArray(normalizedStats.count)
+    },
+    data
+  };
+};

package/test/test.js

@@ -31,6 +31,9 @@ const test = async () => {
 
     const arrObj = indicators.getData()
 
+    const trainSize = Math.round(arrObj.length * 0.8)
+    const testSize = arrObj.length - trainSize
+
     const {
         trainX,
         trainY,
@@ -39,8 +42,10 @@ const test = async () => {
         configX,
         configY
     } = parseTrainingXY({
+        scaling: 'zscore',
         arrObj,
-        trainingSplit: 0.50,
+        trainSize,
+        testSize,
         yCallbackFunc,
         xCallbackFunc,
         validateRows: ({objRow, index}) => {