machine_learning::k_nearest_neighbors::Knn Class Reference

K-Nearest Neighbors (Knn) class using Euclidean distance as distance metric. More...

Collaboration diagram for machine_learning::k_nearest_neighbors::Knn:

Public Member Functions
	Knn (std::vector< std::vector< double > > &X, std::vector< int > &Y)
	Construct a new Knn object.
	Knn (const Knn &model)=default
Knn &	operator= (const Knn &model)=default
	Knn (Knn &&)=default
Knn &	operator= (Knn &&)=default
	~Knn ()=default
	Destroy the Knn object.
int	predict (std::vector< double > &sample, int k)
	Classify sample.

Private Attributes
std::vector< std::vector< double > >	X_ {}
	attributes vector
std::vector< int >	Y_ {}
	labels vector

Detailed Description

K-Nearest Neighbors (Knn) class using Euclidean distance as distance metric.

Constructor & Destructor Documentation

Knn() [1/3]

machine_learning::k_nearest_neighbors::Knn::Knn ( std::vector< std::vector< double > > & X, std::vector< int > & Y )

inlineexplicit

Construct a new Knn object.

Using lazy-learning approch, just holds in memory the dataset.

Parameters

X	attributes vector
Y	labels vector

        : X_(X), Y_(Y){};

Knn() [2/3]

machine_learning::k_nearest_neighbors::Knn::Knn ( const Knn & model )

default

Copy Constructor for class Knn.

Parameters

model

instance of class to be copied

Knn() [3/3]

machine_learning::k_nearest_neighbors::Knn::Knn ( Knn && )

default

Move constructor for class Knn

Member Function Documentation

operator=() [1/2]

Knn & machine_learning::k_nearest_neighbors::Knn::operator= ( const Knn & model )

default

Copy assignment operator for class Knn

operator=() [2/2]

Knn & machine_learning::k_nearest_neighbors::Knn::operator= ( Knn && )

default

Move assignment operator for class Knn

predict()

int machine_learning::k_nearest_neighbors::Knn::predict ( std::vector< double > & sample, int k )

inline

Classify sample.

Parameters

sample	sample
k	number of neighbors

Returns

int label of most frequent neighbors

                                                  {
        std::vector<int> neighbors;
        std::vector<std::pair<double, int>> distances;
        for (size_t i = 0; i < this->X_.size(); ++i) {
            auto current = this->X_.at(i);
            auto label = this->Y_.at(i);
            auto distance = euclidean_distance(current, sample);
            distances.emplace_back(distance, label);
        }
        std::sort(distances.begin(), distances.end());
        for (int i = 0; i < k; i++) {
            auto label = distances.at(i).second;
            neighbors.push_back(label);
        }
        std::unordered_map<int, int> frequency;
        for (auto neighbor : neighbors) {
            ++frequency[neighbor];
        }
        std::pair<int, int> predicted;
        predicted.first = -1;
        predicted.second = -1;
        for (auto& kv : frequency) {
            if (kv.second > predicted.second) {
                predicted.second = kv.second;
                predicted.first = kv.first;
            }
        }
        return predicted.first;
    }

Here is the call graph for this function:

Member Data Documentation

X_

std::vector<std::vector<double> > machine_learning::k_nearest_neighbors::Knn::X_ {}

private

attributes vector

{};  ///< attributes vector

Y_

std::vector<int> machine_learning::k_nearest_neighbors::Knn::Y_ {}

private

labels vector

{};                  ///< labels vector

---

The documentation for this class was generated from the following file:

• machine_learning/k_nearest_neighbors.cpp