KNN.h

Go to the documentation of this file.

 
#include "../Matrix.h"
#include "../matrix_utils.h"
#include "../numerics/utils.h"
#include "Classifier.h"
#include <cstddef>
 
 
class KNN : public Classifier
{
  int nearest_neighbors;
  Matrix<double> trainLabels;
 
public:
  KNN(int neighbors)
    : Classifier()
    , nearest_neighbors(neighbors) { }
 
 
  void fit(const Matrix<double>& X, const Matrix<double>& y) override {
    assert(y.rows() == 1 || y.columns() == 1);
    weights     = X;
    trainLabels = y;
  };
 
  Matrix<double> predict(const Matrix<double>& x) override {
    auto predictions = zerosV(x.rows());
    auto trainX      = weights;
    auto trainY      = trainLabels;
    auto labels      = unique(trainY);
    for(size_t i = 0; i < x.rows(); ++i) {
      auto xi            = x.GetSlice(i);
      auto current_dists = norm(trainX - xi, 0);
      auto k_neares      = trainY.GetSlicesByIndex(argsort(current_dists));
      auto k_nearest     = k_neares.GetSlice(0, nearest_neighbors - 1);
      auto gamma         = zeros(labels.rows(), 1);
      for(size_t k = 0; k < labels.rows(); ++k) {
        std::function<bool(double)> condition = [k](double x) { return bool(x == k); };
        // O(N)
        auto yis = where(condition, k_nearest, { { 1 } }, { { 0 } });
        // O(N)
        gamma(k, 0) = yis.sumElements();
      }
      predictions(i, 0) = argmax(gamma);
    }
    return predictions;
  };
};