//

// This software is released under the MIT License.

// https://opensource.org/licenses/MIT

​

/* ===

ml5 Example

Image classification using MobileNet and p5.js

This example uses a callback pattern to create the classifier

// x: population_cdp

// y: scope1_ghg_emissions_tons_co2e

=== */

let nn;

let data;

​

let counter = 0;

const testInputs = [

  100,

  50000,

  100000,

  500000,

  2500000,

  5000000,

  10000000,

  15000000,

]

​

// Options for Neural Network

const options = {

  inputs: ['population_cdp'],

  outputs: ['scope1_ghg_emissions_tons_co2e'],

  dataUrl: 'data/co2stats.csv',

  task: 'regression',

  debug: true

};

​

​

function setup() {

  createCanvas(400, 400);

  // background(0, 27, 68);

  background(244, 244, 244);

​

  // Step 1: Create Neural Network

  nn = ml5.neuralNetwork(options, modelLoaded);

​

}

​

function modelLoaded() {

  nn.normalizeData();

​

  // visualize the training data

  nn.data.training.forEach(row => {

    const {

      xs,

      ys

    } = row;

    const x = map(xs.population_cdp, 0, 1, 0, width);

    const y = map(ys.scope1_ghg_emissions_tons_co2e, 0, 1, height, 0);

    ellipse(x, y, 2, 2);

  })

​

  const trainingOptions = {

    epochs: 50,

    batchSize: 12

  }

  nn.train(trainingOptions, finishedTraining)

}

​

function finishedTraining() {

​

  if (counter < testInputs.length) {

    const input = testInputs[counter]

    nn.predict([input], (err, result) => {

​

      if (err) {

        console.log(err);

        return;

      }

      console.log(result)

      const {

        inputs,

        outputs

      } = nn.neuralNetworkData.meta;

​

      const inputMin = inputs.population_cdp.min;

      const inputMax = inputs.population_cdp.max;

      const outputMin = outputs.scope1_ghg_emissions_tons_co2e.min;

      const outputMax = outputs.scope1_ghg_emissions_tons_co2e.max;

​

      const x = map(input, inputMin, inputMax, 0, width);

      const y = map(result[0].value, outputMin, outputMax, height, 0);

​

      rectMode(CENTER);

      fill(255, 0, 0);

      text(`pop:${input}`, x, y)

      text(`tons_co2e:${result[0].value}`, x, y + 10)

      rect(x, y, 6, 6);

​

      counter++;

      finishedTraining();

    })

  }

​

}