Machine Learning JS

index.js

const fs = require('fs-extra');
const KNN = require('ml-knn');
const moment = require('moment');
const { GaussianNB, MultinomialNB } = require('ml-naivebayes');
const { Matrix } = require('ml-matrix');
const { DecisionTreeClassifier } = require('ml-cart');




const irisDataset = require('ml-dataset-iris');
// const { DecisionTreeClassifier } = require('ml-cart');

var trainingSet = irisDataset.getNumbers();
var predictions = irisDataset
  .getClasses()
  .map((elem) => irisDataset.getDistinctClasses().indexOf(elem));

var options = {
  gainFunction: 'gini',
  maxDepth: 10,
  minNumSamples: 3
};

var classifier = new DecisionTreeClassifier(options);
classifier.train(trainingSet, predictions);
var result = classifier.predict(trainingSet);

process.exit(0);


const PERCENTAGE_TEST = 0.2;

const loadDataset = (callback) => {

    const filePath = './training-data/poker-hand-training-true.data';

    let dataset = [];
    let labels = [];
    let data = '';

    const lineReader = require('readline').createInterface({
        input: fs.createReadStream(filePath, 'utf8')
    });


    // const readStream = fs.createReadStream(filePath, 'utf8');
    lineReader.on('line', (line) => {
        
        // let arrLines = data.split('\n\r');

        // arrLines.forEach( (val) => {
            let auxData = line.split(',');

            const len = auxData.length;
            if( len < 11 ){
                console.log(auxData);
                process.exit(0);
            }
            for (let i = 0; i < len; i++) {
                auxData[i] = parseInt( auxData[i] );
            }

            if ( isNaN(auxData[0]) ) return;
            
            labels.push(auxData.pop());
            dataset.push(auxData);

        // });

    }).on('close', () => {

        let datasetTest = [];
        let labelsTest = [];

        console.log('[TOTAL ELEMENTS]', dataset.length );
        
        const len = dataset.length;
        for(let i = 0; i < (len*PERCENTAGE_TEST); i++ ){
            datasetTest.push( dataset.pop() );
            labelsTest.push( labels.pop() );
        }

        callback(
            dataset,
            labels,
            datasetTest,
            labelsTest
        );

    });

};

const testResults = (labelsTest, resultTest) => {

    let accurancyCount = 0;
    resultTest.forEach((val, i) => {
        
        // console.log( labelsTest[i], '===>', val );
        if (labelsTest[i] === val) {
            accurancyCount++;
        }

    });

    const lenTest = labelsTest.length;
    console.log( resultTest.length, labelsTest.length );
    return accurancyCount / lenTest;

};

const doKnn = (dataset, labels, datasetTest, labelsTest) => {

    const knn = new KNN(dataset, labels);

    const start = moment();
    console.log(start);

    const resultTest = knn.predict(datasetTest);
    console.log(resultTest);
    console.log(labelsTest);

    console.log( 'ACCURANCY', testResults(labelsTest, resultTest) );

    const end = moment();
    console.log( end.diff(start, 'seconds'), 'seconds' );

};

const doDecisionTree = (dataset, labels, datasetTest, labelsTest) => {

    var options = {
        gainFunction: 'gini',
        maxDepth: 10,
        minNumSamples: 3
    };

    const model = new DecisionTreeClassifier(options);

    const start = moment();
    console.log(start);
    
    console.log('TRAINING...');
    model.train(dataset, labels);
    console.log('TRAINED!');

    const resultTest = model.predict(datasetTest);

    console.log( 'ACCURANCY', testResults(labelsTest, resultTest) );

    const end = moment();
    console.log( end.diff(start, 'seconds'), 'seconds' );

};

const doNaiveBayes_GaussianNB = (dataset, labels, datasetTest, labelsTest) => {

    const model = new GaussianNB();

    let aux = [];
    let auxLabels = [];
    for (let i = 0; i < 5; i++) {
        aux.push( dataset[i] );
        auxLabels.push( Math.random() > .5 ? 1 : 0 );
    }

    const d = new Matrix( aux );
    const l = new Matrix( [auxLabels] );
    
    model.train(d, auxLabels);

    console.log( model );
    return ;

    model.train( aux, auxLabels );

    const resultTest = model.predict([
        [ 1, 2, 1, 4, 1, 5, 1, 3, 1, 6 ]
    ]);

    console.log(resultTest);
    // console.log( 'ACCURANCY', testResults(labelsTest, resultTest) );


};

const doNaiveBayes_MultinomialNB = (dataset, labels, datasetTest, labelsTest) => {

    const model = new MultinomialNB();

    const start = moment();
    console.log(start);

    const datasetMatrix = new Matrix(dataset);

    model.train(datasetMatrix, labels);

    const resultTest = model.predict(datasetTest);

    console.log( 'ACCURANCY', testResults(labelsTest, resultTest) );

    const end = moment();
    console.log( end.diff(start, 'seconds'), 'seconds' );

};

loadDataset( (dataset, labels, datasetTest, labelsTest) => {

    // dataset = [
    //     [1, 1, 0],
    //     [1, 0, 0],
    //     [0, 0, 1],
    //     [0, 1, 1]
    // ];
    // labels = [1, 1, 0, 0];

    // datasetTest = [
    //     [1, 1, 0],
    //     [0, 1, 1],
    //     [1, 0, 1],
    // ];
    // labelsTest = [1,0,0];

    console.log( 'DATASET LENGTH =>', dataset.length);
    console.log( 'LABELS LENGTH =>', labels.length);
    console.log( 'DATASET TEST LENGTH =>', datasetTest.length);
    console.log( 'LABELS TEST LENGTH =>', labelsTest.length);

    doKnn(dataset, labels, datasetTest, labelsTest);
    // doDecisionTree(dataset, labels, datasetTest, labelsTest);
    // doNaiveBayes_MultinomialNB(dataset, labels, datasetTest, labelsTest);

});

// let readStream = fs.createReadStream(filePath, 'utf8');

// readStream.on('data', (chunk) => {
//     data += chunk;
//     console.log(chunk);
// }).on('end', () => { 
//     console.log('end', data);
// });

if (false) {

    const dataset = [
        [0,0,1],
        [0,1,1],
        [0,1,1],
        [1,1,1],
        [1,0,1],
        [0,0,1],
    ];
    const labels = [1,1,0,1,0,0];

    const knn = new KNN(dataset, labels);

    const newDataset = [
        [0,0,0],
        [0,1,0],
    ];

    var ans = knn.predict(newDataset);
    console.log(ans);

}

package.json

{
  "name": "machine-learning-js",
  "version": "1.0.0",
  "description": "",
  "main": "index.js",
  "scripts": {
    "test": "echo \"Error: no test specified\" && exit 1"
  },
  "keywords": [],
  "author": "",
  "license": "ISC",
  "dependencies": {
    "ml-cart": "^1.0.8",
    "ml-knn": "^2.1.3"
  }
}