// http://codingtra.in

// http://patreon.com/codingtrain

​

// Machine Learning

// Nearest Neighbor Recommendations

// More: https://github.com/shiffman/NOC-S17-2-Intelligence-Learning/tree/master/week3-classification-regression

​

// Part 1: https://youtu.be/N8Fabn1om2k

// Part 2: https://youtu.be/Lo89NLmSgl0

// Part 3: https://youtu.be/aMtckmWAzDg

​

// Added styling

​

var data;

var users;

​

var resultP;

var resultDivs = [];

​

function preload() {

  data = loadJSON('movies.json');

}

​

function setup() {

  noCanvas();

  users = {};

​

  var dropdowns = [];

​

  var titles = data.titles;

  for (var i = 0; i < titles.length; i++) {

    var div = createDiv(titles[i]);

    div.position(50, 50 + i * 40);

    var dropdown = createSelect('');

    dropdown.class('dropdown');

    dropdown.position(200, i);

    dropdown.title = titles[i];

    dropdown.option('not seen');

    dropdown.parent(div);

    dropdowns.push(dropdown);

    for (var star = 1; star < 6; star++) {

      dropdown.option(star);

    }

  }

​

​

  var button = createButton('submit');

  button.mousePressed(predictRatings);

  button.position(400, 50);

  resultP = createP('');

  resultP.position(50, 400 );

​

  function predictRatings() {

    var newUser = {};

    for (var i = 0; i < dropdowns.length; i++) {

      var title = dropdowns[i].title;

      var rating = dropdowns[i].value();

      if (rating == 'not seen') {

        rating = null;

      }

      newUser[title] = rating;

    }

​

    findNearestNeighbors(newUser);

  }

​

​

​

​

  function findNearestNeighbors(user) {

    for (var i = 0; i < resultDivs.length; i++) {

      resultDivs[i].remove();

    }

    resultDivs = [];

​

    var similarityScores = {};

    for (var i = 0; i < data.users.length; i++) {

      var other = data.users[i];

      var similarity = euclideanDistance(user, other);

      similarityScores[other.name] = similarity;

    }

​

    data.users.sort(compareSimilarity);

​

    function compareSimilarity(a, b) {

      var score1 = similarityScores[a.name];

      var score2 = similarityScores[b.name];

      return score2 - score1;

    }

​

​

    for (var i = 0; i < data.titles.length; i++) {

      var title = data.titles[i];

      if (user[title] == null) {

​

        var k = 5;

        var weightedSum = 0;

        var similaritySum = 0;

        for (var j = 0; j < k; j++) {

          var name = data.users[j].name;

          var sim = similarityScores[name];

          var ratings = data.users[j];

          var rating = ratings[title];

          if (rating != null) {

            weightedSum += rating * sim;

            similaritySum += sim;

          }

        }

​

        var stars = nf(weightedSum / similaritySum, 1, 2);

        var div = createDiv(title + ': ' + stars);

        resultDivs.push(div);

        div.parent(resultP);

​

​

​

​

​

      }

    }

​

​

​

    // var k = 5;

    // for (var i = 0; i < k; i++) {

    //   var name = data.users[i].name;

    //   var score = nf(similarityScores[name], 1, 2)

    //   var div = createDiv(name + ': ' + score);

    //   resultDivs.push(div);

    //   div.parent(resultP);

    // }

​

​

    // console.log(similarityScores);

​

  }

​

​

​

​

}

​

function euclideanDistance(ratings1, ratings2) {

​

  var titles = data.titles;

​

  var sumSquares = 0;

  for (var i = 0; i < titles.length; i++) {

    var title = titles[i];

    var rating1 = ratings1[title];

    var rating2 = ratings2[title];

    if (rating1 != null && rating2 != null) {

      var diff = rating1 - rating2;

      sumSquares += diff * diff;

    }

  }

  var d = sqrt(sumSquares);

​

  var similarity = 1 / (1 + d);

  return similarity;

}

​