als-statistics/lib/table/instruments/dbscan.js

A powerful and lightweight JavaScript library for descriptive statistics, regression, clustering, outlier detection, and noise analysis using a flexible table/column architecture.

class Dbscan {
   constructor(table, eps = 0.4, minPts = 3) {
      if (table.columnsN < 2) throw new Error('2 or more columns required')
      this.table = table;
      this.eps = eps;
      this.minPts = minPts;
      this.columnsArray = Object.entries(table.columns);
      this.n = this.columnsArray.length;
      this.labels = new Array(this.n).fill(0); // 0: не обработан, -1: шум, 1+: кластер
      this.clusters = [];
      this.distances = null;

      this.computeDistances();
      this.run();
      this.buildClusters();
   }

   computeDistances() { // Вычисление матрицы расстояний
      this.distances = new Array(this.n).fill(null).map(() => new Array(this.n).fill(null));
      for (let i = 0; i < this.n; i++) {
         for (let j = i + 1; j < this.n; j++) {
            const { correlationSample } = this.table.compare(this.columnsArray[i][0], this.columnsArray[j][0]);
            const dist = 1 - correlationSample;
            this.distances[i][j] = dist;
            this.distances[j][i] = dist;
         }
         this.distances[i][i] = 0; // расстояние до себя = 0
      }
   }

   findNeighbors(pointIdx) { // Поиск соседей для точки
      const neighbors = [];
      for (let i = 0; i < this.n; i++) {
         if (this.distances[pointIdx][i] <= this.eps) {
            neighbors.push(i);
         }
      }
      return neighbors;
   }

   expandCluster(pointIdx, clusterId) { // Расширение кластера
      this.labels[pointIdx] = clusterId;
      const seeds = [...this.findNeighbors(pointIdx).filter(idx => idx !== pointIdx)];

      while (seeds.length > 0) {
         const current = seeds.pop();
         if (this.labels[current] === -1) this.labels[current] = clusterId; // шум в кластер
         if (this.labels[current] !== 0) continue; // уже обработан

         this.labels[current] = clusterId;
         const currentNeighbors = this.findNeighbors(current);
         if (currentNeighbors.length >= this.minPts) {
            seeds.push(...currentNeighbors.filter(idx => this.labels[idx] === 0));
         }
      }
   }

   run() { // Основной алгоритм DBSCAN
      let clusterId = 0;
      for (let i = 0; i < this.n; i++) {
         if (this.labels[i] !== 0) continue; // уже обработан

         const neighbors = this.findNeighbors(i);
         if (neighbors.length < this.minPts) {
            this.labels[i] = -1; // шум
            continue;
         }

         clusterId++;
         this.expandCluster(i, clusterId);
      }
   }

   buildClusters() { // Формирование кластеров как таблиц
      for (let id = 1; id <= Math.max(...this.labels); id++) {
         const clusterTable = new this.table.Table();
         this.columnsArray.forEach(([name, column], idx) => {
            if (this.labels[idx] === id) {
               clusterTable.addColumn(name, column.clone(true));
            }
         });
         if (clusterTable.columnsN > 0) this.clusters.push(clusterTable);
      }
   }
}

module.exports = Dbscan