let selectInput; 

let trainBtn; 

let myCanvas; 

let myResults;

let options = {

  inputs: ['x', 'y'],

  outputs: ['label'],

  task: 'classification',

  debug: true

​

}

​

​

function setup() {

  myCanvas = createCanvas(400, 400);

  background(200);

  selectInput = select("#selectInput");

  trainBtn = select("#selectInput");

    myResults = select("#myResults");

  inputValue = selectInput.value();

  selectInput.changed(function(){

  inputValue = selectInput.value();

  });

  neuralnet = ml5.neuralNetwork(options);

  myCanvas.mousePressed(addData);

  trainBtn.mousePressed(trainModel);

}

​

function addData(){

 const xLoc = mouseX;

  const yLoc =  mouseY;

  const labelValue = inputValue;

  neuralnet.addData({x:xLoc,y:yLoc},{label: labelValue})

  ellipse(xLoc, yLoc, 40,40); 

  text(labelValue, xLoc, yLoc);

}

​

function trainModel(){

  neuralnet.normalizeData();

  const trainingOptions = {

    epochs: 24,

    batchSize: 12

  }

  neuralnet.train(trainingOptions, finishedTraining)

​

}

​

function finishedTraining(){

classify();

}

​

function classify(){

neuralnet.classify({x: mouseX, y: mouseY}, gotResults)//if it were a regression you could say prediect

​

}

function gotResults(err,results){

  if(err) {

  console.log(err);

    return; 

  }

  myResults.html("I see a..." + results[0].label);

  classify();

}