Classification-Caner

cancer.js

const tf = require('@tensorflow/tfjs-node');

const TRAINING_DATA_URL = 'https://raw.githubusercontent.com/lmoroney/dlaicourse/master/TensorFlow%20Deployment/Course%201%20-%20TensorFlow-JS/Week%201/Exercise/wdbc-train.csv';
const TESTING_DATA_URL = 'https://raw.githubusercontent.com/lmoroney/dlaicourse/master/TensorFlow%20Deployment/Course%201%20-%20TensorFlow-JS/Week%201/Exercise/wdbc-test.csv';

async function run() {
    // diagnosis column in the labelled csv
    // has all the decisioning data
    const trainingData = tf.data.csv(
        TRAINING_DATA_URL, {
        columnConfigs: {
            diagnosis: {
                isLabel: true
            }
        }
    });

    const testingData = tf.data.csv(
        TESTING_DATA_URL, {
        columnConfigs: {
            diagnosis: {
                isLabel: true
            }
        }
    });


    const numOfFeatures = (await trainingData.columnNames()).length - 1;

    const convertedTrainingData =
        trainingData.map(({ xs, ys }) => {
            const labels = [ys.diagnosis]
            return { xs: Object.values(xs), ys: Object.values(labels) };
        }).batch(100);

    const convertedTestingData =
        testingData.map(({ xs, ys }) => {
            const labels = [ys.diagnosis]
            return { xs: Object.values(xs), ys: Object.values(labels) };
        }).batch(100);


    const model = tf.sequential();
    model.add(tf.layers.dense({ inputShape: [numOfFeatures], activation: "sigmoid", units: numOfFeatures / 2 }))
    model.add(tf.layers.dense({ activation: "softmax", units: 2 }));

    model.compile({ loss: 'categoricalCrossentropy', optimizer: tf.train.adam(0.01) });

    await model.fitDataset(convertedTrainingData,
        {
            epochs: 200,
            validationData: convertedTestingData
        }
    );


    // prediction should be 1
    const testValue1 = tf.tensor2d(
        [0.03044074238, 0.9549225569, 0.1050681432, -0.1227571612, 0.8043023999, 2.735854561, 1.42332992, 0.4536366295, 2.099581594, 1.872171652, -0.4102104418, 1.662683772, -0.3504014478, -0.179655777, 0.3402900665, 6.308376767, 2.739713598, 0.8511109458, 3.737990259, 3.072596581, -0.1056454613, 1.910304965, -0.01866302817, -0.2027993271, 0.9221218488, 4.451638642, 2.915026087, 0.9617363037, 3.594656981, 3.434156634],
        [1, numOfFeatures]
    );

    // prediction should be 0
    const testValue2 = tf.tensor2d(
        [-1.150364822, -0.390641961, -1.128550208, -0.9587635786, 0.3109837048, -0.5959945009, -0.8025961223, -0.8024900198, 0.2945390619, 0.09425149978, -0.4950522998, 1.487201532, -0.5144878222, -0.4915400491, 0.2814983704, -0.6045120643, -0.4690070058, -0.6117000235, 0.05798237404, -0.3576370159, -1.043175598, 0.2135328249, -1.036044595, -0.8488077145, 0.3424985112, -0.7300974341, -0.8123205317, -0.757983665, -0.01614760641, -0.3850340195],
        [1, numOfFeatures]
    );

    const prediction1 = model.predict(testValue1);
    const prediction2 = model.predict(testValue2);


    tf.print(prediction1);
    tf.print(prediction2);

}
run();