@winm2m/inferential-stats-js 0.1.0

package/dist/index.js

@@ -0,0 +1,446 @@
+/**
+ * Encode a UTF-8 string to Uint8Array
+ */
+function encodeString(str) {
+    return new TextEncoder().encode(str);
+}
+/**
+ * Infer column type from values
+ */
+function inferColumnType(values) {
+    for (const v of values) {
+        if (v === null || v === undefined)
+            continue;
+        if (typeof v === 'string')
+            return 'string';
+        if (typeof v === 'number') {
+            if (Number.isInteger(v))
+                continue;
+            return 'float64';
+        }
+    }
+    // Default numeric columns to float64 for compatibility
+    return 'float64';
+}
+/**
+ * Serialize an array of JSON objects into an ArrayBuffer for efficient Worker transfer.
+ * Uses columnar storage with dictionary encoding for strings.
+ *
+ * Format:
+ * [4 bytes: header length][header JSON bytes][column data bytes...]
+ *
+ * Column data layout per column:
+ * - float64/int32: raw typed array bytes
+ * - string: Int32Array of indices into stringTable in header
+ */
+function serializeToBuffer(data) {
+    if (!data || data.length === 0) {
+        // Return minimal buffer for empty data
+        const header = { rowCount: 0, columns: [] };
+        const headerBytes = encodeString(JSON.stringify(header));
+        const buffer = new ArrayBuffer(4 + headerBytes.byteLength);
+        const view = new DataView(buffer);
+        view.setUint32(0, headerBytes.byteLength, true);
+        new Uint8Array(buffer, 4).set(headerBytes);
+        return buffer;
+    }
+    const rowCount = data.length;
+    const columnNames = Object.keys(data[0]);
+    const columns = [];
+    const columnBuffers = [];
+    for (const name of columnNames) {
+        const values = data.map(row => row[name]);
+        const dtype = inferColumnType(values);
+        if (dtype === 'string') {
+            // Dictionary encoding
+            const uniqueValues = [...new Set(values.map(v => String(v ?? '')))];
+            const lookupMap = new Map();
+            uniqueValues.forEach((v, i) => lookupMap.set(v, i));
+            const indices = new Int32Array(rowCount);
+            for (let i = 0; i < rowCount; i++) {
+                indices[i] = lookupMap.get(String(values[i] ?? '')) ?? 0;
+            }
+            columns.push({ name, dtype: 'string', stringTable: uniqueValues });
+            columnBuffers.push(indices.buffer.slice(0));
+        }
+        else {
+            // Numeric column
+            const arr = new Float64Array(rowCount);
+            for (let i = 0; i < rowCount; i++) {
+                const v = values[i];
+                arr[i] = (v === null || v === undefined) ? NaN : Number(v);
+            }
+            columns.push({ name, dtype: 'float64' });
+            columnBuffers.push(arr.buffer.slice(0));
+        }
+    }
+    const header = { rowCount, columns };
+    const headerBytes = encodeString(JSON.stringify(header));
+    // Calculate total size
+    let totalDataSize = 0;
+    for (const buf of columnBuffers) {
+        totalDataSize += buf.byteLength;
+    }
+    const totalSize = 4 + headerBytes.byteLength + totalDataSize;
+    const buffer = new ArrayBuffer(totalSize);
+    const view = new DataView(buffer);
+    // Write header length (4 bytes, little-endian)
+    view.setUint32(0, headerBytes.byteLength, true);
+    // Write header
+    new Uint8Array(buffer, 4, headerBytes.byteLength).set(headerBytes);
+    // Write column data
+    let offset = 4 + headerBytes.byteLength;
+    for (const colBuf of columnBuffers) {
+        new Uint8Array(buffer, offset, colBuf.byteLength).set(new Uint8Array(colBuf));
+        offset += colBuf.byteLength;
+    }
+    return buffer;
+}
+/**
+ * Get a list of transferable ArrayBuffers from the serialized buffer.
+ * Useful for postMessage with Transferable Objects.
+ */
+function getTransferables(buffer) {
+    return [buffer];
+}
+
+/** Custom event name for progress updates */
+const PROGRESS_EVENT_NAME = 'inferential-stats-progress';
+/**
+ * InferentialStats - Main SDK class for browser-based statistical analysis.
+ *
+ * Uses Pyodide (WebAssembly) running in a Web Worker to perform
+ * SPSS-level statistical computations entirely client-side.
+ *
+ * Data is transferred to the worker using ArrayBuffer (Transferable Objects)
+ * for maximum efficiency with large datasets.
+ */
+class InferentialStats {
+    constructor(config = {}) {
+        this.worker = null;
+        this.pendingRequests = new Map();
+        this.requestCounter = 0;
+        this.initialized = false;
+        this.initPromise = null;
+        this.config = {
+            workerUrl: config.workerUrl ?? '',
+            pyodideUrl: config.pyodideUrl ?? 'https://cdn.jsdelivr.net/pyodide/v0.27.5/full/',
+            eventTarget: config.eventTarget ?? (typeof globalThis !== 'undefined' ? globalThis : null),
+        };
+    }
+    /**
+     * Initialize the SDK - loads Pyodide and all required Python packages.
+     * Must be called before any analysis methods.
+     * Dispatches 'inferential-stats-progress' CustomEvents during loading.
+     */
+    async init() {
+        if (this.initialized)
+            return;
+        if (this.initPromise)
+            return this.initPromise;
+        this.initPromise = this._doInit();
+        return this.initPromise;
+    }
+    async _doInit() {
+        // Create worker
+        if (this.config.workerUrl) {
+            this.worker = new Worker(this.config.workerUrl);
+        }
+        else {
+            throw new Error('workerUrl is required. Point it to the stats-worker.js file from @winm2m/inferential-stats-js/worker');
+        }
+        // Set up message handler
+        this.worker.onmessage = (event) => {
+            this._handleWorkerMessage(event.data);
+        };
+        this.worker.onerror = (event) => {
+            console.error('[InferentialStats] Worker error:', event.message);
+        };
+        // Wait for worker ready signal
+        await new Promise((resolve) => {
+            const readyHandler = (event) => {
+                if (event.data.id === '__worker_ready__') {
+                    resolve();
+                }
+            };
+            this.worker.addEventListener('message', readyHandler, { once: true });
+        });
+        // Initialize Pyodide in the worker
+        await this._sendRequest('init', undefined, {
+            pyodideUrl: this.config.pyodideUrl,
+        });
+        this.initialized = true;
+    }
+    /** Handle messages from the worker */
+    _handleWorkerMessage(response) {
+        if (response.type === 'progress' && response.progress) {
+            this._dispatchProgress(response.progress);
+            return; // Don't resolve the promise for progress messages
+        }
+        const pending = this.pendingRequests.get(response.id);
+        if (!pending)
+            return;
+        this.pendingRequests.delete(response.id);
+        if (response.type === 'error') {
+            pending.reject(new Error(response.error ?? 'Unknown worker error'));
+        }
+        else {
+            pending.resolve(response.data);
+        }
+    }
+    /** Dispatch a progress CustomEvent */
+    _dispatchProgress(detail) {
+        try {
+            if (this.config.eventTarget && typeof CustomEvent !== 'undefined') {
+                const event = new CustomEvent(PROGRESS_EVENT_NAME, { detail });
+                this.config.eventTarget.dispatchEvent(event);
+            }
+        }
+        catch {
+            // Silently ignore if CustomEvent is not available (e.g., Node.js test env)
+        }
+    }
+    /**
+     * Send a request to the worker and return a promise for the result.
+     * Optionally serializes data to ArrayBuffer for efficient transfer.
+     */
+    _sendRequest(type, data, params) {
+        if (!this.worker) {
+            return Promise.reject(new Error('Worker not initialized. Call init() first.'));
+        }
+        const id = `req_${++this.requestCounter}_${Date.now()}`;
+        return new Promise((resolve, reject) => {
+            this.pendingRequests.set(id, {
+                resolve: resolve,
+                reject,
+            });
+            const request = { id, type, params };
+            if (data && data.length > 0) {
+                // Serialize to ArrayBuffer for efficient transfer
+                const buffer = serializeToBuffer(data);
+                request.payload = buffer;
+                this.worker.postMessage(request, getTransferables(buffer));
+            }
+            else {
+                this.worker.postMessage(request);
+            }
+        });
+    }
+    /** Helper to wrap analysis calls with timing */
+    async _runAnalysis(type, input) {
+        if (!this.initialized) {
+            throw new Error('SDK not initialized. Call init() first.');
+        }
+        const startTime = performance.now();
+        try {
+            const { data, ...params } = input;
+            const result = await this._sendRequest(type, data, params);
+            return {
+                success: true,
+                data: result,
+                executionTimeMs: Math.round(performance.now() - startTime),
+            };
+        }
+        catch (err) {
+            return {
+                success: false,
+                data: null,
+                error: err instanceof Error ? err.message : String(err),
+                executionTimeMs: Math.round(performance.now() - startTime),
+            };
+        }
+    }
+    // ═══════════════════════════════════════
+    // ① Descriptive Statistics
+    // ═══════════════════════════════════════
+    /** Frequency analysis for categorical variables */
+    async frequencies(input) {
+        return this._runAnalysis('frequencies', input);
+    }
+    /** Descriptive statistics for continuous variables */
+    async descriptives(input) {
+        return this._runAnalysis('descriptives', input);
+    }
+    /** Cross-tabulation with Chi-square test */
+    async crosstabs(input) {
+        return this._runAnalysis('crosstabs', input);
+    }
+    // ═══════════════════════════════════════
+    // ② Compare Means
+    // ═══════════════════════════════════════
+    /** Independent-samples T-test with Levene's test */
+    async ttestIndependent(input) {
+        return this._runAnalysis('ttest_independent', input);
+    }
+    /** Paired-samples T-test */
+    async ttestPaired(input) {
+        return this._runAnalysis('ttest_paired', input);
+    }
+    /** One-Way ANOVA */
+    async anovaOneway(input) {
+        return this._runAnalysis('anova_oneway', input);
+    }
+    /** Post-hoc Tukey HSD test */
+    async posthocTukey(input) {
+        return this._runAnalysis('posthoc_tukey', input);
+    }
+    // ═══════════════════════════════════════
+    // ③ Regression
+    // ═══════════════════════════════════════
+    /** Linear regression (OLS) */
+    async linearRegression(input) {
+        return this._runAnalysis('linear_regression', input);
+    }
+    /** Binary logistic regression */
+    async logisticBinary(input) {
+        return this._runAnalysis('logistic_binary', input);
+    }
+    /** Multinomial logistic regression */
+    async logisticMultinomial(input) {
+        return this._runAnalysis('logistic_multinomial', input);
+    }
+    // ═══════════════════════════════════════
+    // ④ Classify
+    // ═══════════════════════════════════════
+    /** K-Means clustering */
+    async kmeans(input) {
+        return this._runAnalysis('kmeans', input);
+    }
+    /** Hierarchical (agglomerative) clustering */
+    async hierarchicalCluster(input) {
+        return this._runAnalysis('hierarchical_cluster', input);
+    }
+    // ═══════════════════════════════════════
+    // ⑤ Dimension Reduction
+    // ═══════════════════════════════════════
+    /** Exploratory Factor Analysis (EFA) with varimax rotation */
+    async efa(input) {
+        return this._runAnalysis('efa', input);
+    }
+    /** Principal Component Analysis (PCA) */
+    async pca(input) {
+        return this._runAnalysis('pca', input);
+    }
+    /** Multidimensional Scaling (MDS) */
+    async mds(input) {
+        return this._runAnalysis('mds', input);
+    }
+    // ═══════════════════════════════════════
+    // ⑥ Scale
+    // ═══════════════════════════════════════
+    /** Reliability analysis (Cronbach's Alpha) */
+    async cronbachAlpha(input) {
+        return this._runAnalysis('cronbach_alpha', input);
+    }
+    // ═══════════════════════════════════════
+    // Lifecycle
+    // ═══════════════════════════════════════
+    /** Check if the SDK is initialized */
+    isInitialized() {
+        return this.initialized;
+    }
+    /** Terminate the worker and clean up resources */
+    destroy() {
+        if (this.worker) {
+            this.worker.terminate();
+            this.worker = null;
+        }
+        this.initialized = false;
+        this.initPromise = null;
+        this.pendingRequests.clear();
+    }
+}
+
+/**
+ * Decode UTF-8 bytes to string
+ */
+function decodeString(bytes) {
+    return new TextDecoder().decode(bytes);
+}
+/**
+ * Deserialize an ArrayBuffer (created by serializeToBuffer) back to JSON objects.
+ */
+function deserializeFromBuffer(buffer) {
+    const view = new DataView(buffer);
+    // Read header length
+    const headerLength = view.getUint32(0, true);
+    // Read header
+    const headerBytes = new Uint8Array(buffer, 4, headerLength);
+    const header = JSON.parse(decodeString(headerBytes));
+    if (header.rowCount === 0)
+        return [];
+    const { rowCount, columns } = header;
+    let offset = 4 + headerLength;
+    // Read column data
+    const columnData = new Map();
+    for (const col of columns) {
+        if (col.dtype === 'string') {
+            const byteLen = rowCount * 4; // Int32 = 4 bytes
+            const aligned = new Int32Array(new Uint8Array(buffer, offset, byteLen).slice().buffer);
+            const values = [];
+            for (let i = 0; i < rowCount; i++) {
+                values.push(col.stringTable[aligned[i]]);
+            }
+            columnData.set(col.name, values);
+            offset += byteLen;
+        }
+        else {
+            const byteLen = rowCount * 8; // Float64 = 8 bytes
+            const aligned = new Float64Array(new Uint8Array(buffer, offset, byteLen).slice().buffer);
+            const values = [];
+            for (let i = 0; i < rowCount; i++) {
+                values.push(aligned[i]);
+            }
+            columnData.set(col.name, values);
+            offset += byteLen;
+        }
+    }
+    // Reconstruct row-oriented JSON
+    const result = [];
+    for (let i = 0; i < rowCount; i++) {
+        const row = {};
+        for (const col of columns) {
+            row[col.name] = columnData.get(col.name)[i];
+        }
+        result.push(row);
+    }
+    return result;
+}
+/**
+ * Deserialize buffer and convert to column-oriented format.
+ * More efficient for statistical computations.
+ */
+function deserializeToColumns(buffer) {
+    const view = new DataView(buffer);
+    const headerLength = view.getUint32(0, true);
+    const headerBytes = new Uint8Array(buffer, 4, headerLength);
+    const header = JSON.parse(decodeString(headerBytes));
+    if (header.rowCount === 0)
+        return { rowCount: 0, columns: {} };
+    const { rowCount, columns: colMeta } = header;
+    let offset = 4 + headerLength;
+    const columns = {};
+    for (const col of colMeta) {
+        if (col.dtype === 'string') {
+            const byteLen = rowCount * 4;
+            const indices = new Int32Array(new Uint8Array(buffer, offset, byteLen).slice().buffer);
+            const values = [];
+            for (let i = 0; i < rowCount; i++) {
+                values.push(col.stringTable[indices[i]]);
+            }
+            columns[col.name] = { dtype: 'string', values };
+            offset += byteLen;
+        }
+        else {
+            const byteLen = rowCount * 8;
+            const arr = new Float64Array(new Uint8Array(buffer, offset, byteLen).slice().buffer);
+            const values = Array.from(arr);
+            columns[col.name] = { dtype: 'float64', values };
+            offset += byteLen;
+        }
+    }
+    return { rowCount, columns };
+}
+
+export { InferentialStats, PROGRESS_EVENT_NAME, deserializeFromBuffer, deserializeToColumns, getTransferables, serializeToBuffer };
+//# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sources":["../src/bridge/serializer.ts","../src/InferentialStats.ts","../src/bridge/deserializer.ts"],"sourcesContent":[null,null,null],"names":[],"mappings":"AAEA;;AAEG;AACH,SAAS,YAAY,CAAC,GAAW,EAAA;IAC/B,OAAO,IAAI,WAAW,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC;AACtC;AAEA;;AAEG;AACH,SAAS,eAAe,CAAC,MAAiB,EAAA;AACxC,IAAA,KAAK,MAAM,CAAC,IAAI,MAAM,EAAE;AACtB,QAAA,IAAI,CAAC,KAAK,IAAI,IAAI,CAAC,KAAK,SAAS;YAAE;QACnC,IAAI,OAAO,CAAC,KAAK,QAAQ;AAAE,YAAA,OAAO,QAAQ;AAC1C,QAAA,IAAI,OAAO,CAAC,KAAK,QAAQ,EAAE;AACzB,YAAA,IAAI,MAAM,CAAC,SAAS,CAAC,CAAC,CAAC;gBAAE;AACzB,YAAA,OAAO,SAAS;QAClB;IACF;;AAEA,IAAA,OAAO,SAAS;AAClB;AAEA;;;;;;;;;;AAUG;AACG,SAAU,iBAAiB,CAAC,IAA+B,EAAA;IAC/D,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE;;QAE9B,MAAM,MAAM,GAAsB,EAAE,QAAQ,EAAE,CAAC,EAAE,OAAO,EAAE,EAAE,EAAE;QAC9D,MAAM,WAAW,GAAG,YAAY,CAAC,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;QACxD,MAAM,MAAM,GAAG,IAAI,WAAW,CAAC,CAAC,GAAG,WAAW,CAAC,UAAU,CAAC;AAC1D,QAAA,MAAM,IAAI,GAAG,IAAI,QAAQ,CAAC,MAAM,CAAC;QACjC,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,WAAW,CAAC,UAAU,EAAE,IAAI,CAAC;QAC/C,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,WAAW,CAAC;AAC1C,QAAA,OAAO,MAAM;IACf;AAEA,IAAA,MAAM,QAAQ,GAAG,IAAI,CAAC,MAAM;IAC5B,MAAM,WAAW,GAAG,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IACxC,MAAM,OAAO,GAAiB,EAAE;IAChC,MAAM,aAAa,GAAkB,EAAE;AAEvC,IAAA,KAAK,MAAM,IAAI,IAAI,WAAW,EAAE;AAC9B,QAAA,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,IAAI,GAAG,CAAC,IAAI,CAAC,CAAC;AACzC,QAAA,MAAM,KAAK,GAAG,eAAe,CAAC,MAAM,CAAC;AAErC,QAAA,IAAI,KAAK,KAAK,QAAQ,EAAE;;YAEtB,MAAM,YAAY,GAAG,CAAC,GAAG,IAAI,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,IAAI,MAAM,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC;AACnE,YAAA,MAAM,SAAS,GAAG,IAAI,GAAG,EAAkB;AAC3C,YAAA,YAAY,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,KAAK,SAAS,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;AAEnD,YAAA,MAAM,OAAO,GAAG,IAAI,UAAU,CAAC,QAAQ,CAAC;AACxC,YAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE;gBACjC,OAAO,CAAC,CAAC,CAAC,GAAG,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC;YAC1D;AAEA,YAAA,OAAO,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,QAAQ,EAAE,WAAW,EAAE,YAAY,EAAE,CAAC;AAClE,YAAA,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;QAC7C;aAAO;;AAEL,YAAA,MAAM,GAAG,GAAG,IAAI,YAAY,CAAC,QAAQ,CAAC;AACtC,YAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE;AACjC,gBAAA,MAAM,CAAC,GAAG,MAAM,CAAC,CAAC,CAAC;gBACnB,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,IAAI,IAAI,CAAC,KAAK,SAAS,IAAI,GAAG,GAAG,MAAM,CAAC,CAAC,CAAC;YAC5D;YACA,OAAO,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,SAAS,EAAE,CAAC;AACxC,YAAA,aAAa,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;QACzC;IACF;AAEA,IAAA,MAAM,MAAM,GAAsB,EAAE,QAAQ,EAAE,OAAO,EAAE;IACvD,MAAM,WAAW,GAAG,YAAY,CAAC,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;;IAGxD,IAAI,aAAa,GAAG,CAAC;AACrB,IAAA,KAAK,MAAM,GAAG,IAAI,aAAa,EAAE;AAC/B,QAAA,aAAa,IAAI,GAAG,CAAC,UAAU;IACjC;IAEA,MAAM,SAAS,GAAG,CAAC,GAAG,WAAW,CAAC,UAAU,GAAG,aAAa;AAC5D,IAAA,MAAM,MAAM,GAAG,IAAI,WAAW,CAAC,SAAS,CAAC;AACzC,IAAA,MAAM,IAAI,GAAG,IAAI,QAAQ,CAAC,MAAM,CAAC;;IAGjC,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,WAAW,CAAC,UAAU,EAAE,IAAI,CAAC;;AAG/C,IAAA,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC,EAAE,WAAW,CAAC,UAAU,CAAC,CAAC,GAAG,CAAC,WAAW,CAAC;;AAGlE,IAAA,IAAI,MAAM,GAAG,CAAC,GAAG,WAAW,CAAC,UAAU;AACvC,IAAA,KAAK,MAAM,MAAM,IAAI,aAAa,EAAE;AAClC,QAAA,IAAI,UAAU,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC,UAAU,CAAC,CAAC,GAAG,CAAC,IAAI,UAAU,CAAC,MAAM,CAAC,CAAC;AAC7E,QAAA,MAAM,IAAI,MAAM,CAAC,UAAU;IAC7B;AAEA,IAAA,OAAO,MAAM;AACf;AAEA;;;AAGG;AACG,SAAU,gBAAgB,CAAC,MAAmB,EAAA;IAClD,OAAO,CAAC,MAAM,CAAC;AACjB;;AC1EA;AACO,MAAM,mBAAmB,GAAG;AAEnC;;;;;;;;AAQG;MACU,gBAAgB,CAAA;AAW3B,IAAA,WAAA,CAAY,SAAiC,EAAE,EAAA;QAVvC,IAAA,CAAA,MAAM,GAAkB,IAAI;AAC5B,QAAA,IAAA,CAAA,eAAe,GAGlB,IAAI,GAAG,EAAE;QACN,IAAA,CAAA,cAAc,GAAG,CAAC;QAElB,IAAA,CAAA,WAAW,GAAG,KAAK;QACnB,IAAA,CAAA,WAAW,GAAyB,IAAI;QAG9C,IAAI,CAAC,MAAM,GAAG;AACZ,YAAA,SAAS,EAAE,MAAM,CAAC,SAAS,IAAI,EAAE;AACjC,YAAA,UAAU,EAAE,MAAM,CAAC,UAAU,IAAI,gDAAgD;AACjF,YAAA,WAAW,EAAE,MAAM,CAAC,WAAW,KAAK,OAAO,UAAU,KAAK,WAAW,GAAG,UAAU,GAAG,IAA8B,CAAC;SACrH;IACH;AAEA;;;;AAIG;AACH,IAAA,MAAM,IAAI,GAAA;QACR,IAAI,IAAI,CAAC,WAAW;YAAE;QACtB,IAAI,IAAI,CAAC,WAAW;YAAE,OAAO,IAAI,CAAC,WAAW;AAE7C,QAAA,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,OAAO,EAAE;QACjC,OAAO,IAAI,CAAC,WAAW;IACzB;AAEQ,IAAA,MAAM,OAAO,GAAA;;AAEnB,QAAA,IAAI,IAAI,CAAC,MAAM,CAAC,SAAS,EAAE;AACzB,YAAA,IAAI,CAAC,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC;QACjD;aAAO;AACL,YAAA,MAAM,IAAI,KAAK,CACb,sGAAsG,CACvG;QACH;;QAGA,IAAI,CAAC,MAAM,CAAC,SAAS,GAAG,CAAC,KAAmC,KAAI;AAC9D,YAAA,IAAI,CAAC,oBAAoB,CAAC,KAAK,CAAC,IAAI,CAAC;AACvC,QAAA,CAAC;QAED,IAAI,CAAC,MAAM,CAAC,OAAO,GAAG,CAAC,KAAiB,KAAI;YAC1C,OAAO,CAAC,KAAK,CAAC,kCAAkC,EAAE,KAAK,CAAC,OAAO,CAAC;AAClE,QAAA,CAAC;;AAGD,QAAA,MAAM,IAAI,OAAO,CAAO,CAAC,OAAO,KAAI;AAClC,YAAA,MAAM,YAAY,GAAG,CAAC,KAAmC,KAAI;gBAC3D,IAAI,KAAK,CAAC,IAAI,CAAC,EAAE,KAAK,kBAAkB,EAAE;AACxC,oBAAA,OAAO,EAAE;gBACX;AACF,YAAA,CAAC;AACD,YAAA,IAAI,CAAC,MAAO,CAAC,gBAAgB,CAAC,SAAS,EAAE,YAAY,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;AACxE,QAAA,CAAC,CAAC;;AAGF,QAAA,MAAM,IAAI,CAAC,YAAY,CAA2B,MAAM,EAAE,SAAS,EAAE;AACnE,YAAA,UAAU,EAAE,IAAI,CAAC,MAAM,CAAC,UAAU;AACnC,SAAA,CAAC;AAEF,QAAA,IAAI,CAAC,WAAW,GAAG,IAAI;IACzB;;AAGQ,IAAA,oBAAoB,CAAC,QAAwB,EAAA;QACnD,IAAI,QAAQ,CAAC,IAAI,KAAK,UAAU,IAAI,QAAQ,CAAC,QAAQ,EAAE;AACrD,YAAA,IAAI,CAAC,iBAAiB,CAAC,QAAQ,CAAC,QAAQ,CAAC;AACzC,YAAA,OAAO;QACT;AAEA,QAAA,MAAM,OAAO,GAAG,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAE,CAAC;AACrD,QAAA,IAAI,CAAC,OAAO;YAAE;QAEd,IAAI,CAAC,eAAe,CAAC,MAAM,CAAC,QAAQ,CAAC,EAAE,CAAC;AAExC,QAAA,IAAI,QAAQ,CAAC,IAAI,KAAK,OAAO,EAAE;AAC7B,YAAA,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,QAAQ,CAAC,KAAK,IAAI,sBAAsB,CAAC,CAAC;QACrE;aAAO;AACL,YAAA,OAAO,CAAC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC;QAChC;IACF;;AAGQ,IAAA,iBAAiB,CAAC,MAAsB,EAAA;AAC9C,QAAA,IAAI;YACF,IAAI,IAAI,CAAC,MAAM,CAAC,WAAW,IAAI,OAAO,WAAW,KAAK,WAAW,EAAE;gBACjE,MAAM,KAAK,GAAG,IAAI,WAAW,CAAC,mBAAmB,EAAE,EAAE,MAAM,EAAE,CAAC;gBAC9D,IAAI,CAAC,MAAM,CAAC,WAAW,CAAC,aAAa,CAAC,KAAK,CAAC;YAC9C;QACF;AAAE,QAAA,MAAM;;QAER;IACF;AAEA;;;AAGG;AACK,IAAA,YAAY,CAClB,IAAuB,EACvB,IAAgC,EAChC,MAAgC,EAAA;AAEhC,QAAA,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE;YAChB,OAAO,OAAO,CAAC,MAAM,CAAC,IAAI,KAAK,CAAC,4CAA4C,CAAC,CAAC;QAChF;AAEA,QAAA,MAAM,EAAE,GAAG,CAAA,IAAA,EAAO,EAAE,IAAI,CAAC,cAAc,CAAA,CAAA,EAAI,IAAI,CAAC,GAAG,EAAE,EAAE;QAEvD,OAAO,IAAI,OAAO,CAAI,CAAC,OAAO,EAAE,MAAM,KAAI;AACxC,YAAA,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,EAAE,EAAE;AAC3B,gBAAA,OAAO,EAAE,OAAmC;gBAC5C,MAAM;AACP,aAAA,CAAC;YAEF,MAAM,OAAO,GAAkB,EAAE,EAAE,EAAE,IAAI,EAAE,MAAM,EAAE;YAEnD,IAAI,IAAI,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE;;AAE3B,gBAAA,MAAM,MAAM,GAAG,iBAAiB,CAAC,IAAI,CAAC;AACtC,gBAAA,OAAO,CAAC,OAAO,GAAG,MAAM;AACxB,gBAAA,IAAI,CAAC,MAAO,CAAC,WAAW,CAAC,OAAO,EAAE,gBAAgB,CAAC,MAAM,CAAC,CAAC;YAC7D;iBAAO;AACL,gBAAA,IAAI,CAAC,MAAO,CAAC,WAAW,CAAC,OAAO,CAAC;YACnC;AACF,QAAA,CAAC,CAAC;IACJ;;AAGQ,IAAA,MAAM,YAAY,CACxB,IAAuB,EACvB,KAAQ,EAAA;AAER,QAAA,IAAI,CAAC,IAAI,CAAC,WAAW,EAAE;AACrB,YAAA,MAAM,IAAI,KAAK,CAAC,yCAAyC,CAAC;QAC5D;AAEA,QAAA,MAAM,SAAS,GAAG,WAAW,CAAC,GAAG,EAAE;AACnC,QAAA,IAAI;YACF,MAAM,EAAE,IAAI,EAAE,GAAG,MAAM,EAAE,GAAG,KAAK;AACjC,YAAA,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,YAAY,CAAI,IAAI,EAAE,IAAI,EAAE,MAAiC,CAAC;YACxF,OAAO;AACL,gBAAA,OAAO,EAAE,IAAI;AACb,gBAAA,IAAI,EAAE,MAAM;gBACZ,eAAe,EAAE,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,GAAG,EAAE,GAAG,SAAS,CAAC;aAC3D;QACH;QAAE,OAAO,GAAG,EAAE;YACZ,OAAO;AACL,gBAAA,OAAO,EAAE,KAAK;AACd,gBAAA,IAAI,EAAE,IAAoB;AAC1B,gBAAA,KAAK,EAAE,GAAG,YAAY,KAAK,GAAG,GAAG,CAAC,OAAO,GAAG,MAAM,CAAC,GAAG,CAAC;gBACvD,eAAe,EAAE,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,GAAG,EAAE,GAAG,SAAS,CAAC;aAC3D;QACH;IACF;;;;;IAOA,MAAM,WAAW,CAAC,KAAuB,EAAA;QACvC,OAAO,IAAI,CAAC,YAAY,CAAC,aAAa,EAAE,KAAK,CAAC;IAChD;;IAGA,MAAM,YAAY,CAAC,KAAwB,EAAA;QACzC,OAAO,IAAI,CAAC,YAAY,CAAC,cAAc,EAAE,KAAK,CAAC;IACjD;;IAGA,MAAM,SAAS,CAAC,KAAqB,EAAA;QACnC,OAAO,IAAI,CAAC,YAAY,CAAC,WAAW,EAAE,KAAK,CAAC;IAC9C;;;;;IAOA,MAAM,gBAAgB,CAAC,KAA4B,EAAA;QACjD,OAAO,IAAI,CAAC,YAAY,CAAC,mBAAmB,EAAE,KAAK,CAAC;IACtD;;IAGA,MAAM,WAAW,CAAC,KAAuB,EAAA;QACvC,OAAO,IAAI,CAAC,YAAY,CAAC,cAAc,EAAE,KAAK,CAAC;IACjD;;IAGA,MAAM,WAAW,CAAC,KAAiB,EAAA;QACjC,OAAO,IAAI,CAAC,YAAY,CAAC,cAAc,EAAE,KAAK,CAAC;IACjD;;IAGA,MAAM,YAAY,CAAC,KAAmB,EAAA;QACpC,OAAO,IAAI,CAAC,YAAY,CAAC,eAAe,EAAE,KAAK,CAAC;IAClD;;;;;IAOA,MAAM,gBAAgB,CAAC,KAA4B,EAAA;QACjD,OAAO,IAAI,CAAC,YAAY,CAAC,mBAAmB,EAAE,KAAK,CAAC;IACtD;;IAGA,MAAM,cAAc,CAAC,KAA0B,EAAA;QAC7C,OAAO,IAAI,CAAC,YAAY,CAAC,iBAAiB,EAAE,KAAK,CAAC;IACpD;;IAGA,MAAM,mBAAmB,CAAC,KAA+B,EAAA;QACvD,OAAO,IAAI,CAAC,YAAY,CAAC,sBAAsB,EAAE,KAAK,CAAC;IACzD;;;;;IAOA,MAAM,MAAM,CAAC,KAAkB,EAAA;QAC7B,OAAO,IAAI,CAAC,YAAY,CAAC,QAAQ,EAAE,KAAK,CAAC;IAC3C;;IAGA,MAAM,mBAAmB,CAAC,KAA+B,EAAA;QACvD,OAAO,IAAI,CAAC,YAAY,CAAC,sBAAsB,EAAE,KAAK,CAAC;IACzD;;;;;IAOA,MAAM,GAAG,CAAC,KAAe,EAAA;QACvB,OAAO,IAAI,CAAC,YAAY,CAAC,KAAK,EAAE,KAAK,CAAC;IACxC;;IAGA,MAAM,GAAG,CAAC,KAAe,EAAA;QACvB,OAAO,IAAI,CAAC,YAAY,CAAC,KAAK,EAAE,KAAK,CAAC;IACxC;;IAGA,MAAM,GAAG,CAAC,KAAe,EAAA;QACvB,OAAO,IAAI,CAAC,YAAY,CAAC,KAAK,EAAE,KAAK,CAAC;IACxC;;;;;IAOA,MAAM,aAAa,CAAC,KAAyB,EAAA;QAC3C,OAAO,IAAI,CAAC,YAAY,CAAC,gBAAgB,EAAE,KAAK,CAAC;IACnD;;;;;IAOA,aAAa,GAAA;QACX,OAAO,IAAI,CAAC,WAAW;IACzB;;IAGA,OAAO,GAAA;AACL,QAAA,IAAI,IAAI,CAAC,MAAM,EAAE;AACf,YAAA,IAAI,CAAC,MAAM,CAAC,SAAS,EAAE;AACvB,YAAA,IAAI,CAAC,MAAM,GAAG,IAAI;QACpB;AACA,QAAA,IAAI,CAAC,WAAW,GAAG,KAAK;AACxB,QAAA,IAAI,CAAC,WAAW,GAAG,IAAI;AACvB,QAAA,IAAI,CAAC,eAAe,CAAC,KAAK,EAAE;IAC9B;AACD;;AClVD;;AAEG;AACH,SAAS,YAAY,CAAC,KAAiB,EAAA;IACrC,OAAO,IAAI,WAAW,EAAE,CAAC,MAAM,CAAC,KAAK,CAAC;AACxC;AAEA;;AAEG;AACG,SAAU,qBAAqB,CAAC,MAAmB,EAAA;AACvD,IAAA,MAAM,IAAI,GAAG,IAAI,QAAQ,CAAC,MAAM,CAAC;;IAGjC,MAAM,YAAY,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,CAAC;;IAG5C,MAAM,WAAW,GAAG,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC,EAAE,YAAY,CAAC;IAC3D,MAAM,MAAM,GAAsB,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,WAAW,CAAC,CAAC;AAEvE,IAAA,IAAI,MAAM,CAAC,QAAQ,KAAK,CAAC;AAAE,QAAA,OAAO,EAAE;AAEpC,IAAA,MAAM,EAAE,QAAQ,EAAE,OAAO,EAAE,GAAG,MAAM;AACpC,IAAA,IAAI,MAAM,GAAG,CAAC,GAAG,YAAY;;AAG7B,IAAA,MAAM,UAAU,GAAqC,IAAI,GAAG,EAAE;AAE9D,IAAA,KAAK,MAAM,GAAG,IAAI,OAAO,EAAE;AACzB,QAAA,IAAI,GAAG,CAAC,KAAK,KAAK,QAAQ,EAAE;AAC1B,YAAA,MAAM,OAAO,GAAG,QAAQ,GAAG,CAAC,CAAC;YAC7B,MAAM,OAAO,GAAG,IAAI,UAAU,CAAC,IAAI,UAAU,CAAC,MAAM,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC,KAAK,EAAE,CAAC,MAAM,CAAC;YACtF,MAAM,MAAM,GAAa,EAAE;AAC3B,YAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE;AACjC,gBAAA,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,WAAY,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3C;YACA,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,EAAE,MAAM,CAAC;YAChC,MAAM,IAAI,OAAO;QACnB;aAAO;AACL,YAAA,MAAM,OAAO,GAAG,QAAQ,GAAG,CAAC,CAAC;YAC7B,MAAM,OAAO,GAAG,IAAI,YAAY,CAAC,IAAI,UAAU,CAAC,MAAM,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC,KAAK,EAAE,CAAC,MAAM,CAAC;YACxF,MAAM,MAAM,GAAa,EAAE;AAC3B,YAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE;gBACjC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;YACzB;YACA,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,EAAE,MAAM,CAAC;YAChC,MAAM,IAAI,OAAO;QACnB;IACF;;IAGA,MAAM,MAAM,GAA8B,EAAE;AAC5C,IAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE;QACjC,MAAM,GAAG,GAA4B,EAAE;AACvC,QAAA,KAAK,MAAM,GAAG,IAAI,OAAO,EAAE;AACzB,YAAA,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAE,CAAC,CAAC,CAAC;QAC9C;AACA,QAAA,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC;IAClB;AAEA,IAAA,OAAO,MAAM;AACf;AAEA;;;AAGG;AACG,SAAU,oBAAoB,CAAC,MAAmB,EAAA;AAItD,IAAA,MAAM,IAAI,GAAG,IAAI,QAAQ,CAAC,MAAM,CAAC;IACjC,MAAM,YAAY,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,CAAC;IAC5C,MAAM,WAAW,GAAG,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC,EAAE,YAAY,CAAC;IAC3D,MAAM,MAAM,GAAsB,IAAI,CAAC,KAAK,CAAC,YAAY,CAAC,WAAW,CAAC,CAAC;AAEvE,IAAA,IAAI,MAAM,CAAC,QAAQ,KAAK,CAAC;QAAE,OAAO,EAAE,QAAQ,EAAE,CAAC,EAAE,OAAO,EAAE,EAAE,EAAE;IAE9D,MAAM,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,EAAE,GAAG,MAAM;AAC7C,IAAA,IAAI,MAAM,GAAG,CAAC,GAAG,YAAY;IAC7B,MAAM,OAAO,GAAmE,EAAE;AAElF,IAAA,KAAK,MAAM,GAAG,IAAI,OAAO,EAAE;AACzB,QAAA,IAAI,GAAG,CAAC,KAAK,KAAK,QAAQ,EAAE;AAC1B,YAAA,MAAM,OAAO,GAAG,QAAQ,GAAG,CAAC;YAC5B,MAAM,OAAO,GAAG,IAAI,UAAU,CAAC,IAAI,UAAU,CAAC,MAAM,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC,KAAK,EAAE,CAAC,MAAM,CAAC;YACtF,MAAM,MAAM,GAAa,EAAE;AAC3B,YAAA,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE;AACjC,gBAAA,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,WAAY,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC;YAC3C;AACA,YAAA,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,EAAE,KAAK,EAAE,QAAQ,EAAE,MAAM,EAAE;YAC/C,MAAM,IAAI,OAAO;QACnB;aAAO;AACL,YAAA,MAAM,OAAO,GAAG,QAAQ,GAAG,CAAC;YAC5B,MAAM,GAAG,GAAG,IAAI,YAAY,CAAC,IAAI,UAAU,CAAC,MAAM,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC,KAAK,EAAE,CAAC,MAAM,CAAC;YACpF,MAAM,MAAM,GAAa,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC;AACxC,YAAA,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,EAAE,KAAK,EAAE,SAAS,EAAE,MAAM,EAAE;YAChD,MAAM,IAAI,OAAO;QACnB;IACF;AAEA,IAAA,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE;AAC9B;;;;"}

package/dist/python/classify.d.ts

@@ -0,0 +1,5 @@
+/**
+ * Python code for classification/clustering functions.
+ */
+export declare const KMEANS_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\nfrom sklearn.cluster import KMeans\nfrom sklearn.preprocessing import StandardScaler\n\ndef run_kmeans(data_json, variables_json, k, max_iterations=300, random_state=42):\n    df = pd.DataFrame(json.loads(data_json))\n    variables = json.loads(variables_json)\n    \n    X = df[variables].apply(pd.to_numeric, errors='coerce').dropna()\n    \n    scaler = StandardScaler()\n    X_scaled = scaler.fit_transform(X)\n    \n    model = KMeans(n_clusters=k, max_iter=max_iterations, random_state=random_state, n_init=10)\n    labels = model.fit_predict(X_scaled)\n    \n    # Transform centers back to original scale\n    centers_original = scaler.inverse_transform(model.cluster_centers_)\n    \n    centers = []\n    for i in range(k):\n        center = {}\n        for j, var in enumerate(variables):\n            center[var] = round(float(centers_original[i, j]), 6)\n        centers.append({'cluster': i, 'center': center})\n    \n    unique, counts = np.unique(labels, return_counts=True)\n    cluster_sizes = {int(u): int(c) for u, c in zip(unique, counts)}\n    \n    return json.dumps({\n        'labels': [int(l) for l in labels],\n        'centers': centers,\n        'inertia': round(float(model.inertia_), 6),\n        'iterations': int(model.n_iter_),\n        'clusterSizes': cluster_sizes\n    })\n";
+export declare const HIERARCHICAL_CLUSTER_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\nfrom scipy.cluster.hierarchy import linkage, fcluster, dendrogram\nfrom sklearn.preprocessing import StandardScaler\n\ndef run_hierarchical_cluster(data_json, variables_json, method='ward', metric='euclidean', n_clusters=None, distance_threshold=None):\n    df = pd.DataFrame(json.loads(data_json))\n    variables = json.loads(variables_json)\n    \n    X = df[variables].apply(pd.to_numeric, errors='coerce').dropna()\n    \n    scaler = StandardScaler()\n    X_scaled = scaler.fit_transform(X)\n    \n    Z = linkage(X_scaled, method=method, metric=metric)\n    \n    if n_clusters is not None:\n        labels = fcluster(Z, t=n_clusters, criterion='maxclust')\n    elif distance_threshold is not None:\n        labels = fcluster(Z, t=distance_threshold, criterion='distance')\n    else:\n        labels = fcluster(Z, t=3, criterion='maxclust')\n    \n    labels = labels - 1  # 0-indexed\n    \n    unique, counts = np.unique(labels, return_counts=True)\n    cluster_sizes = {int(u): int(c) for u, c in zip(unique, counts)}\n    \n    # Dendrogram data (truncated for large datasets)\n    trunc = min(30, len(X_scaled))\n    dend = dendrogram(Z, truncate_mode='lastp', p=trunc, no_plot=True)\n    \n    return json.dumps({\n        'labels': [int(l) for l in labels],\n        'nClusters': len(unique),\n        'linkageMatrix': [[round(float(x), 6) for x in row] for row in Z.tolist()],\n        'clusterSizes': cluster_sizes,\n        'dendrogramData': {\n            'icoord': [[round(float(x), 4) for x in row] for row in dend['icoord']],\n            'dcoord': [[round(float(x), 4) for x in row] for row in dend['dcoord']],\n            'leaves': [int(x) for x in dend['leaves']]\n        }\n    })\n";

package/dist/python/compare-means.d.ts

@@ -0,0 +1,7 @@
+/**
+ * Python code for compare means functions.
+ */
+export declare const TTEST_INDEPENDENT_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\nfrom scipy import stats\n\ndef run_ttest_independent(data_json, variable, group_variable, group1_value, group2_value):\n    df = pd.DataFrame(json.loads(data_json))\n    \n    g1 = pd.to_numeric(df[df[group_variable] == group1_value][variable], errors='coerce').dropna()\n    g2 = pd.to_numeric(df[df[group_variable] == group2_value][variable], errors='coerce').dropna()\n    \n    # Levene's test for equality of variances\n    levene_stat, levene_p = stats.levene(g1, g2)\n    equal_var = levene_p > 0.05\n    \n    # T-test with equal variance\n    t_eq, p_eq = stats.ttest_ind(g1, g2, equal_var=True)\n    # T-test with unequal variance (Welch's)\n    t_uneq, p_uneq = stats.ttest_ind(g1, g2, equal_var=False)\n    \n    mean_diff = float(g1.mean() - g2.mean())\n    \n    # Degrees of freedom\n    df_eq = len(g1) + len(g2) - 2\n    \n    # Welch df\n    s1_sq = g1.var(ddof=1)\n    s2_sq = g2.var(ddof=1)\n    n1, n2 = len(g1), len(g2)\n    num = (s1_sq/n1 + s2_sq/n2)**2\n    denom = (s1_sq/n1)**2/(n1-1) + (s2_sq/n2)**2/(n2-1)\n    df_welch = float(num/denom)\n    \n    # Confidence intervals\n    se_eq = float(np.sqrt(((n1-1)*s1_sq + (n2-1)*s2_sq)/(n1+n2-2) * (1/n1 + 1/n2)))\n    se_uneq = float(np.sqrt(s1_sq/n1 + s2_sq/n2))\n    \n    ci_eq = stats.t.interval(0.95, df_eq, loc=mean_diff, scale=se_eq)\n    ci_uneq = stats.t.interval(0.95, df_welch, loc=mean_diff, scale=se_uneq)\n    \n    def make_result(t_stat, df_val, p_val, ci):\n        return {\n            'tStatistic': round(float(t_stat), 6),\n            'degreesOfFreedom': round(float(df_val), 6),\n            'pValue': float(p_val),\n            'meanDifference': round(mean_diff, 6),\n            'confidenceInterval': [round(float(ci[0]), 6), round(float(ci[1]), 6)],\n            'group1Mean': round(float(g1.mean()), 6),\n            'group1Std': round(float(g1.std(ddof=1)), 6),\n            'group1N': n1,\n            'group2Mean': round(float(g2.mean()), 6),\n            'group2Std': round(float(g2.std(ddof=1)), 6),\n            'group2N': n2\n        }\n    \n    return json.dumps({\n        'leveneTest': {\n            'statistic': round(float(levene_stat), 6),\n            'pValue': float(levene_p),\n            'equalVariance': bool(equal_var)\n        },\n        'equalVariance': make_result(t_eq, df_eq, p_eq, ci_eq),\n        'unequalVariance': make_result(t_uneq, df_welch, p_uneq, ci_uneq)\n    })\n";
+export declare const TTEST_PAIRED_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\nfrom scipy import stats\n\ndef run_ttest_paired(data_json, variable1, variable2):\n    df = pd.DataFrame(json.loads(data_json))\n    \n    v1 = pd.to_numeric(df[variable1], errors='coerce').dropna()\n    v2 = pd.to_numeric(df[variable2], errors='coerce').dropna()\n    \n    # Align by index\n    common = v1.index.intersection(v2.index)\n    v1 = v1.loc[common]\n    v2 = v2.loc[common]\n    \n    diff = v1 - v2\n    n = len(diff)\n    \n    t_stat, p_val = stats.ttest_rel(v1, v2)\n    \n    mean_diff = float(diff.mean())\n    std_diff = float(diff.std(ddof=1))\n    se = std_diff / np.sqrt(n)\n    ci = stats.t.interval(0.95, n-1, loc=mean_diff, scale=se)\n    \n    return json.dumps({\n        'tStatistic': round(float(t_stat), 6),\n        'degreesOfFreedom': n - 1,\n        'pValue': float(p_val),\n        'meanDifference': round(mean_diff, 6),\n        'stdDifference': round(std_diff, 6),\n        'confidenceInterval': [round(float(ci[0]), 6), round(float(ci[1]), 6)],\n        'mean1': round(float(v1.mean()), 6),\n        'mean2': round(float(v2.mean()), 6),\n        'n': n\n    })\n";
+export declare const ANOVA_ONEWAY_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\nfrom scipy import stats\n\ndef run_anova_oneway(data_json, variable, group_variable):\n    df = pd.DataFrame(json.loads(data_json))\n    \n    groups = df.groupby(group_variable)[variable].apply(\n        lambda x: pd.to_numeric(x, errors='coerce').dropna().tolist()\n    )\n    \n    group_arrays = [np.array(g) for g in groups.values if len(g) > 0]\n    group_names = [str(name) for name, g in zip(groups.index, groups.values) if len(g) > 0]\n    \n    f_stat, p_val = stats.f_oneway(*group_arrays)\n    \n    # Compute detailed ANOVA table\n    grand_mean = np.concatenate(group_arrays).mean()\n    n_total = sum(len(g) for g in group_arrays)\n    k = len(group_arrays)\n    \n    ss_between = sum(len(g) * (g.mean() - grand_mean)**2 for g in group_arrays)\n    ss_within = sum(((g - g.mean())**2).sum() for g in group_arrays)\n    \n    df_between = k - 1\n    df_within = n_total - k\n    \n    ms_between = ss_between / df_between\n    ms_within = ss_within / df_within\n    \n    ss_total = ss_between + ss_within\n    eta_sq = ss_between / ss_total if ss_total > 0 else 0\n    \n    group_stats = []\n    for name, arr in zip(group_names, group_arrays):\n        group_stats.append({\n            'group': name,\n            'n': len(arr),\n            'mean': round(float(arr.mean()), 6),\n            'std': round(float(arr.std(ddof=1)), 6)\n        })\n    \n    return json.dumps({\n        'fStatistic': round(float(f_stat), 6),\n        'pValue': float(p_val),\n        'degreesOfFreedomBetween': df_between,\n        'degreesOfFreedomWithin': df_within,\n        'sumOfSquaresBetween': round(float(ss_between), 6),\n        'sumOfSquaresWithin': round(float(ss_within), 6),\n        'meanSquareBetween': round(float(ms_between), 6),\n        'meanSquareWithin': round(float(ms_within), 6),\n        'groupStats': group_stats,\n        'etaSquared': round(float(eta_sq), 6)\n    })\n";
+export declare const POSTHOC_TUKEY_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\nfrom statsmodels.stats.multicomp import pairwise_tukeyhsd\n\ndef run_posthoc_tukey(data_json, variable, group_variable, alpha=0.05):\n    df = pd.DataFrame(json.loads(data_json))\n    \n    df[variable] = pd.to_numeric(df[variable], errors='coerce')\n    df = df.dropna(subset=[variable])\n    \n    result = pairwise_tukeyhsd(df[variable], df[group_variable], alpha=alpha)\n    \n    comparisons = []\n    for i in range(len(result.summary().data) - 1):\n        row = result.summary().data[i + 1]\n        comparisons.append({\n            'group1': str(row[0]),\n            'group2': str(row[1]),\n            'meanDifference': round(float(row[2]), 6),\n            'pValue': round(float(row[3]), 6),\n            'lowerCI': round(float(row[4]), 6),\n            'upperCI': round(float(row[5]), 6),\n            'reject': bool(row[6])\n        })\n    \n    return json.dumps({\n        'comparisons': comparisons,\n        'alpha': alpha\n    })\n";

package/dist/python/descriptive.d.ts

@@ -0,0 +1,7 @@
+/**
+ * Python code for descriptive statistics functions.
+ * These are executed inside Pyodide in the Web Worker.
+ */
+export declare const FREQUENCIES_PY = "\nimport json\nimport pandas as pd\n\ndef run_frequencies(data_json, variable):\n    df = pd.DataFrame(json.loads(data_json))\n    series = df[variable]\n    total = len(series)\n    \n    counts = series.value_counts(dropna=False)\n    pcts = series.value_counts(normalize=True, dropna=False) * 100\n    \n    freqs = []\n    cum_pct = 0\n    for val in counts.index:\n        count = int(counts[val])\n        pct = float(pcts[val])\n        cum_pct += pct\n        freqs.append({\n            'value': str(val) if not isinstance(val, (int, float)) else val,\n            'count': count,\n            'percentage': round(pct, 4),\n            'cumulativePercentage': round(cum_pct, 4)\n        })\n    \n    return json.dumps({\n        'variable': variable,\n        'totalCount': total,\n        'frequencies': freqs\n    })\n";
+export declare const DESCRIPTIVES_PY = "\nimport json\nimport pandas as pd\nfrom scipy import stats as sp_stats\n\ndef run_descriptives(data_json, variables_json):\n    df = pd.DataFrame(json.loads(data_json))\n    variables = json.loads(variables_json)\n    \n    results = []\n    for var in variables:\n        col = pd.to_numeric(df[var], errors='coerce').dropna()\n        desc = col.describe()\n        results.append({\n            'variable': var,\n            'count': int(desc['count']),\n            'mean': round(float(desc['mean']), 6),\n            'std': round(float(desc['std']), 6),\n            'min': round(float(desc['min']), 6),\n            'max': round(float(desc['max']), 6),\n            'q25': round(float(desc['25%']), 6),\n            'q50': round(float(desc['50%']), 6),\n            'q75': round(float(desc['75%']), 6),\n            'skewness': round(float(sp_stats.skew(col)), 6),\n            'kurtosis': round(float(sp_stats.kurtosis(col)), 6)\n        })\n    \n    return json.dumps({'statistics': results})\n";
+export declare const CROSSTABS_PY = "\nimport json\nimport pandas as pd\nfrom scipy.stats import chi2_contingency\nimport numpy as np\n\ndef run_crosstabs(data_json, row_variable, col_variable):\n    df = pd.DataFrame(json.loads(data_json))\n    \n    ct = pd.crosstab(df[row_variable], df[col_variable])\n    chi2, p, dof, expected = chi2_contingency(ct)\n    \n    n = ct.values.sum()\n    k = min(ct.shape) - 1\n    cramers_v = float(np.sqrt(chi2 / (n * k))) if k > 0 else 0\n    \n    row_labels = [str(x) for x in ct.index.tolist()]\n    col_labels = [str(x) for x in ct.columns.tolist()]\n    \n    table = []\n    row_sums = ct.sum(axis=1)\n    col_sums = ct.sum(axis=0)\n    total = ct.values.sum()\n    \n    for i, rl in enumerate(row_labels):\n        for j, cl in enumerate(col_labels):\n            obs = int(ct.iloc[i, j])\n            exp = float(expected[i, j])\n            table.append({\n                'row': rl,\n                'col': cl,\n                'observed': obs,\n                'expected': round(exp, 4),\n                'rowPercentage': round(obs / float(row_sums.iloc[i]) * 100, 4) if row_sums.iloc[i] > 0 else 0,\n                'colPercentage': round(obs / float(col_sums.iloc[j]) * 100, 4) if col_sums.iloc[j] > 0 else 0,\n                'totalPercentage': round(obs / float(total) * 100, 4) if total > 0 else 0\n            })\n    \n    return json.dumps({\n        'rowVariable': row_variable,\n        'colVariable': col_variable,\n        'table': table,\n        'rowLabels': row_labels,\n        'colLabels': col_labels,\n        'chiSquare': round(float(chi2), 6),\n        'degreesOfFreedom': int(dof),\n        'pValue': float(p),\n        'cramersV': round(cramers_v, 6)\n    })\n";

package/dist/python/dimension.d.ts

@@ -0,0 +1,6 @@
+/**
+ * Python code for dimension reduction functions.
+ */
+export declare const EFA_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\nfrom factor_analyzer import FactorAnalyzer\nfrom factor_analyzer.factor_analyzer import calculate_bartlett_sphericity, calculate_kmo\n\ndef run_efa(data_json, variables_json, n_factors, rotation='varimax', method='minres'):\n    df = pd.DataFrame(json.loads(data_json))\n    variables = json.loads(variables_json)\n    \n    X = df[variables].apply(pd.to_numeric, errors='coerce').dropna()\n    \n    # KMO and Bartlett tests\n    kmo_all, kmo_model = calculate_kmo(X)\n    chi2, p_value = calculate_bartlett_sphericity(X)\n    \n    fa = FactorAnalyzer(n_factors=n_factors, rotation=rotation, method=method)\n    fa.fit(X)\n    \n    loadings = fa.loadings_\n    loadings_dict = {}\n    for i, var in enumerate(variables):\n        loadings_dict[var] = [round(float(x), 6) for x in loadings[i]]\n    \n    ev, v = fa.get_factor_variance()\n    \n    communalities = fa.get_communalities()\n    uniquenesses = fa.get_uniquenesses()\n    \n    comm_dict = {}\n    uniq_dict = {}\n    for i, var in enumerate(variables):\n        comm_dict[var] = round(float(communalities[i]), 6)\n        uniq_dict[var] = round(float(uniquenesses[i]), 6)\n    \n    eigenvalues = fa.get_eigenvalues()[0]\n    \n    return json.dumps({\n        'loadings': loadings_dict,\n        'eigenvalues': [round(float(x), 6) for x in eigenvalues],\n        'variance': [round(float(x), 6) for x in ev],\n        'cumulativeVariance': [round(float(sum(v[:i+1])), 6) for i in range(len(v))],\n        'communalities': comm_dict,\n        'uniquenesses': uniq_dict,\n        'nFactors': n_factors,\n        'rotation': rotation,\n        'kmo': round(float(kmo_model), 6),\n        'bartlettChi2': round(float(chi2), 6),\n        'bartlettPValue': float(p_value)\n    })\n";
+export declare const PCA_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\nfrom sklearn.decomposition import PCA\nfrom sklearn.preprocessing import StandardScaler\n\ndef run_pca(data_json, variables_json, n_components=None, standardize=True):\n    df = pd.DataFrame(json.loads(data_json))\n    variables = json.loads(variables_json)\n    \n    X = df[variables].apply(pd.to_numeric, errors='coerce').dropna()\n    \n    if standardize:\n        scaler = StandardScaler()\n        X_scaled = scaler.fit_transform(X)\n    else:\n        X_scaled = X.values\n    \n    if n_components is None:\n        n_components = min(len(variables), len(X_scaled))\n    \n    pca = PCA(n_components=n_components)\n    transformed = pca.fit_transform(X_scaled)\n    \n    loadings = {}\n    for i, var in enumerate(variables):\n        loadings[var] = [round(float(x), 6) for x in pca.components_[:, i]]\n    \n    cum_var = np.cumsum(pca.explained_variance_ratio_)\n    \n    return json.dumps({\n        'components': [[round(float(x), 6) for x in row] for row in transformed.tolist()],\n        'explainedVariance': [round(float(x), 6) for x in pca.explained_variance_],\n        'explainedVarianceRatio': [round(float(x), 6) for x in pca.explained_variance_ratio_],\n        'cumulativeVarianceRatio': [round(float(x), 6) for x in cum_var],\n        'loadings': loadings,\n        'singularValues': [round(float(x), 6) for x in pca.singular_values_],\n        'nComponents': n_components\n    })\n";
+export declare const MDS_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\nfrom sklearn.manifold import MDS\nfrom sklearn.preprocessing import StandardScaler\n\ndef run_mds(data_json, variables_json, n_components=2, metric=True, max_iterations=300, random_state=42):\n    df = pd.DataFrame(json.loads(data_json))\n    variables = json.loads(variables_json)\n    \n    X = df[variables].apply(pd.to_numeric, errors='coerce').dropna()\n    \n    scaler = StandardScaler()\n    X_scaled = scaler.fit_transform(X)\n    \n    mds = MDS(n_components=n_components, metric=metric, max_iter=max_iterations, random_state=random_state, normalized_stress='auto')\n    coords = mds.fit_transform(X_scaled)\n    \n    return json.dumps({\n        'coordinates': [[round(float(x), 6) for x in row] for row in coords.tolist()],\n        'stress': round(float(mds.stress_), 6),\n        'nComponents': n_components\n    })\n";

package/dist/python/index.d.ts

@@ -0,0 +1,6 @@
+export { FREQUENCIES_PY, DESCRIPTIVES_PY, CROSSTABS_PY } from './descriptive';
+export { TTEST_INDEPENDENT_PY, TTEST_PAIRED_PY, ANOVA_ONEWAY_PY, POSTHOC_TUKEY_PY } from './compare-means';
+export { LINEAR_REGRESSION_PY, LOGISTIC_BINARY_PY, LOGISTIC_MULTINOMIAL_PY } from './regression';
+export { KMEANS_PY, HIERARCHICAL_CLUSTER_PY } from './classify';
+export { EFA_PY, PCA_PY, MDS_PY } from './dimension';
+export { CRONBACH_ALPHA_PY } from './scale';

package/dist/python/regression.d.ts

@@ -0,0 +1,6 @@
+/**
+ * Python code for regression analysis functions.
+ */
+export declare const LINEAR_REGRESSION_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\nimport statsmodels.api as sm\n\ndef run_linear_regression(data_json, dependent, independents_json, add_constant=True):\n    df = pd.DataFrame(json.loads(data_json))\n    independents = json.loads(independents_json)\n    \n    y = pd.to_numeric(df[dependent], errors='coerce')\n    X = df[independents].apply(pd.to_numeric, errors='coerce')\n    \n    mask = y.notna() & X.notna().all(axis=1)\n    y = y[mask]\n    X = X[mask]\n    \n    if add_constant:\n        X = sm.add_constant(X)\n    \n    model = sm.OLS(y, X).fit()\n    \n    coefficients = []\n    for i, name in enumerate(model.params.index):\n        ci = model.conf_int().iloc[i]\n        coefficients.append({\n            'variable': str(name),\n            'coefficient': round(float(model.params.iloc[i]), 6),\n            'stdError': round(float(model.bse.iloc[i]), 6),\n            'tStatistic': round(float(model.tvalues.iloc[i]), 6),\n            'pValue': float(model.pvalues.iloc[i]),\n            'confidenceInterval': [round(float(ci[0]), 6), round(float(ci[1]), 6)]\n        })\n    \n    dw = float(sm.stats.stattools.durbin_watson(model.resid))\n    \n    return json.dumps({\n        'rSquared': round(float(model.rsquared), 6),\n        'adjustedRSquared': round(float(model.rsquared_adj), 6),\n        'fStatistic': round(float(model.fvalue), 6),\n        'fPValue': float(model.f_pvalue),\n        'coefficients': coefficients,\n        'residualStdError': round(float(np.sqrt(model.mse_resid)), 6),\n        'observations': int(model.nobs),\n        'degreesOfFreedom': int(model.df_resid),\n        'durbinWatson': round(dw, 6)\n    })\n";
+export declare const LOGISTIC_BINARY_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\nimport statsmodels.api as sm\n\ndef run_logistic_binary(data_json, dependent, independents_json, add_constant=True):\n    df = pd.DataFrame(json.loads(data_json))\n    independents = json.loads(independents_json)\n    \n    y = pd.to_numeric(df[dependent], errors='coerce')\n    X = df[independents].apply(pd.to_numeric, errors='coerce')\n    \n    mask = y.notna() & X.notna().all(axis=1)\n    y = y[mask]\n    X = X[mask]\n    \n    if add_constant:\n        X = sm.add_constant(X)\n    \n    model = sm.Logit(y, X).fit(disp=0)\n    \n    coefficients = []\n    ci = model.conf_int()\n    for i, name in enumerate(model.params.index):\n        coef = float(model.params.iloc[i])\n        coefficients.append({\n            'variable': str(name),\n            'coefficient': round(coef, 6),\n            'stdError': round(float(model.bse.iloc[i]), 6),\n            'zStatistic': round(float(model.tvalues.iloc[i]), 6),\n            'pValue': float(model.pvalues.iloc[i]),\n            'oddsRatio': round(float(np.exp(coef)), 6),\n            'confidenceInterval': [round(float(ci.iloc[i, 0]), 6), round(float(ci.iloc[i, 1]), 6)]\n        })\n    \n    return json.dumps({\n        'coefficients': coefficients,\n        'pseudoRSquared': round(float(model.prsquared), 6),\n        'logLikelihood': round(float(model.llf), 6),\n        'llrPValue': float(model.llr_pvalue),\n        'aic': round(float(model.aic), 6),\n        'bic': round(float(model.bic), 6),\n        'observations': int(model.nobs),\n        'convergence': bool(model.mle_retvals['converged'])\n    })\n";
+export declare const LOGISTIC_MULTINOMIAL_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\nimport statsmodels.api as sm\n\ndef run_logistic_multinomial(data_json, dependent, independents_json, reference_category=None):\n    df = pd.DataFrame(json.loads(data_json))\n    independents = json.loads(independents_json)\n    \n    X = df[independents].apply(pd.to_numeric, errors='coerce')\n    y = df[dependent]\n    \n    mask = X.notna().all(axis=1) & y.notna()\n    X = X[mask]\n    y = y[mask]\n    \n    # Encode categories\n    categories = sorted(y.unique().tolist(), key=str)\n    if reference_category is not None:\n        ref = str(reference_category)\n    else:\n        ref = str(categories[0])\n    \n    y_coded = pd.Categorical(y, categories=categories)\n    y_dummies = pd.get_dummies(y_coded, drop_first=False)\n    \n    X_const = sm.add_constant(X)\n    \n    from sklearn.linear_model import LogisticRegression\n    \n    le_map = {str(c): i for i, c in enumerate(categories)}\n    y_numeric = y.map(lambda x: le_map[str(x)])\n    \n    model = LogisticRegression(multi_class='multinomial', solver='lbfgs', max_iter=1000)\n    model.fit(X, y_numeric)\n    \n    ref_idx = le_map[ref]\n    non_ref_cats = [c for c in categories if str(c) != ref]\n    \n    coefficients = []\n    for cat in non_ref_cats:\n        cat_idx = le_map[str(cat)]\n        for j, var_name in enumerate(independents):\n            coef = float(model.coef_[cat_idx][j] - model.coef_[ref_idx][j])\n            coefficients.append({\n                'category': str(cat),\n                'variable': var_name,\n                'coefficient': round(coef, 6),\n                'stdError': 0.0,\n                'zStatistic': 0.0,\n                'pValue': 0.0,\n                'oddsRatio': round(float(np.exp(coef)), 6),\n                'confidenceInterval': [0.0, 0.0]\n            })\n    \n    # Log-likelihood\n    proba = model.predict_proba(X)\n    ll = float(np.sum(np.log(proba[np.arange(len(y_numeric)), y_numeric] + 1e-10)))\n    n_params = len(non_ref_cats) * (len(independents) + 1)\n    aic = -2 * ll + 2 * n_params\n    bic_val = -2 * ll + n_params * np.log(len(y))\n    \n    return json.dumps({\n        'coefficients': coefficients,\n        'pseudoRSquared': round(float(model.score(X, y_numeric)), 6),\n        'logLikelihood': round(ll, 6),\n        'llrPValue': 0.0,\n        'aic': round(float(aic), 6),\n        'bic': round(float(bic_val), 6),\n        'categories': [str(c) for c in categories],\n        'referenceCategory': ref,\n        'observations': int(len(y))\n    })\n";

package/dist/python/scale.d.ts

@@ -0,0 +1,4 @@
+/**
+ * Python code for scale/reliability functions.
+ */
+export declare const CRONBACH_ALPHA_PY = "\nimport json\nimport pandas as pd\nimport numpy as np\n\ndef run_cronbach_alpha(data_json, items_json):\n    df = pd.DataFrame(json.loads(data_json))\n    items = json.loads(items_json)\n    \n    X = df[items].apply(pd.to_numeric, errors='coerce').dropna()\n    \n    n_items = len(items)\n    n_obs = len(X)\n    \n    # Compute Cronbach's Alpha\n    item_vars = X.var(ddof=1)\n    total_var = X.sum(axis=1).var(ddof=1)\n    alpha = (n_items / (n_items - 1)) * (1 - item_vars.sum() / total_var)\n    \n    # Standardized alpha (using correlation matrix)\n    corr_matrix = X.corr()\n    mean_r = (corr_matrix.sum().sum() - n_items) / (n_items * (n_items - 1))\n    std_alpha = (n_items * mean_r) / (1 + (n_items - 1) * mean_r)\n    \n    # Item analysis\n    item_analysis = []\n    total_score = X.sum(axis=1)\n    \n    for item in items:\n        item_col = X[item]\n        other_items = [i for i in items if i != item]\n        other_sum = X[other_items].sum(axis=1)\n        \n        # Corrected item-total correlation\n        citc = float(item_col.corr(other_sum))\n        \n        # Alpha if item deleted\n        if len(other_items) > 1:\n            sub_X = X[other_items]\n            sub_vars = sub_X.var(ddof=1)\n            sub_total_var = sub_X.sum(axis=1).var(ddof=1)\n            k = len(other_items)\n            alpha_deleted = (k / (k - 1)) * (1 - sub_vars.sum() / sub_total_var)\n        else:\n            alpha_deleted = 0.0\n        \n        item_analysis.append({\n            'item': item,\n            'itemMean': round(float(item_col.mean()), 6),\n            'itemStd': round(float(item_col.std(ddof=1)), 6),\n            'correctedItemTotalCorrelation': round(citc, 6),\n            'alphaIfItemDeleted': round(float(alpha_deleted), 6)\n        })\n    \n    return json.dumps({\n        'alpha': round(float(alpha), 6),\n        'standardizedAlpha': round(float(std_alpha), 6),\n        'nItems': n_items,\n        'nObservations': n_obs,\n        'itemAnalysis': item_analysis,\n        'interItemCorrelationMean': round(float(mean_r), 6)\n    })\n";