Untitled

require('@tensorflow/tfjs-node');   //running tfjs on laptop cpu
const tf = require('@tensorflow/tfjs'); //require in Tensorflow JS lib
const  Users = require('../../mongodb/models/users')
const Posts = require('../../mongodb/models/posts')

function knn(features, labels, predictionPoint, k) { 
features = tf.tensor(features);
labels = tf.tensor(labels);
    const {mean, variance} = tf.moments(features, 0);
    const scaledPrediction = predictionPoint.sub(mean).div(variance.pow(0.5))    //Scaled Prediction const for standardization

return (
   features
/* Step 0 - Standarization of features for  */
    .sub(mean)
    .div(variance.pow(0.5))
/* Step 1 - Find Distance between features & prediction point features */
    .sub(scaledPrediction)	//Broadcasting operation
    .pow(2)	//Elementwise operation to square each element
    .sum(1) //Sums along x(1)-axis
    .sqrt()	//Elementwise operation to take power of .5 = sqrt of each element
/* Step 2 - Sort from lowest distance to greatest distance */
    .expandDims(1) //We expand dimensions of distances tensor across the x-axis to get the shape of [4,1], the same as the labels distance
    .concat(labels, 1) //We concatenate labels to distances across the x-axis so they are linked by the same indices in one tensor
    .unstack() //We unstack our 1 tensor into 1 Vanilla JS array containing multiple tensors
/* After unstacking our tensor, we are dealing with vanilla JS array and we can ONLY use vanilla JS methods from this point on */
    .sort((a,b) => a.get(0) > b.get(0) ? 1 : -1) //Sorting function tp sprt tensors in order of least to greatest distance
    /* Step 3 - Average the label value of the top k records */ 
    .slice(0, k)//Get Top k records
)}
module.exports = knn