gpmp::ml::KNN Class Reference

Represents a K Nearest Neighbors (KNN) classifier. More...

#include <knn.hpp>

Public Member Functions
	KNN ()
	Constructor for the KNN class. More...
	~KNN ()
	Destructor for the KNN class. More...
void	train (const std::vector< std::vector< double >> &training_data, const std::vector< int > &labels)
	Trains the KNN model with the given training data and labels. More...
int	predict (const std::vector< double > &input_vector, int k)
	Predicts the label of a given input vector using KNN algorithm. More...

Private Member Functions
double	calculateEuclideanDistance (const std::vector< double > &vec1, const std::vector< double > &vec2)
	Calculates the Euclidean distance between two vectors. More...

Private Attributes
std::vector< std::vector< double > >	training_data
std::vector< int >	labels

Detailed Description

Represents a K Nearest Neighbors (KNN) classifier.

Definition at line 52 of file knn.hpp.

Constructor & Destructor Documentation

KNN()

gpmp::ml::KNN::KNN

(

)

Constructor for the KNN class.

Definition at line 39 of file knn.cpp.

               {
}

~KNN()

gpmp::ml::KNN::~KNN

(

)

Destructor for the KNN class.

Definition at line 42 of file knn.cpp.

                {
}

Member Function Documentation

calculateEuclideanDistance()

double gpmp::ml::KNN::calculateEuclideanDistance ( const std::vector< double > &  vec1, const std::vector< double > &  vec2  )

private

Calculates the Euclidean distance between two vectors.

Parameters

vec1	The first vector
vec2	The second vector

Returns

The Euclidean distance between the two vectors

Definition at line 107 of file knn.cpp.

                                                                         {
    double distance = 0.0;
    for (size_t i = 0; i < vec1.size(); ++i) {
        double diff = vec1[i] - vec2[i];
        distance += diff * diff;
    }
    return std::sqrt(distance);
}

predict()

int gpmp::ml::KNN::predict	(	const std::vector< double > &	input_vector,
		int	k
	)

Predicts the label of a given input vector using KNN algorithm.

Parameters

input_vector	The input vector for which prediction is to be made
k	The number of nearest neighbors to consider

Returns

The predicted label

Definition at line 56 of file knn.cpp.

                                                                   {
    if (training_data.empty() || labels.empty()) {
        throw std::logic_error(
            "Model not trained. Call train() before predict.");
    }
 
    if (input_vector.size() != training_data[0].size()) {
        throw std::invalid_argument("Invalid input vector size.");
    }
 
    // if (k <= 0 || k > training_data.size()) {
    if (k <= 0 || static_cast<size_t>(k) > training_data.size()) {
        throw std::invalid_argument("Invalid value of k.");
    }
 
    // Calculate distances and store index-label pairs
    std::vector<std::pair<double, int>> distances;
    for (size_t i = 0; i < training_data.size(); ++i) {
        double distance =
            calculateEuclideanDistance(input_vector, training_data[i]);
        distances.emplace_back(distance, labels[i]);
    }
 
    // Sort distances in ascending order
    std::sort(
        distances.begin(),
        distances.end(),
        [](const std::pair<double, int> &a, const std::pair<double, int> &b) {
            return a.first < b.first;
        });
 
    // Count votes for each label among the k nearest neighbors
    std::unordered_map<int, int> label_counts;
    for (int i = 0; i < k; ++i) {
        label_counts[distances[i].second]++;
    }
 
    // Find the label with the maximum votes
    int max_votes = -1;
    int predicted_label = -1;
    for (const auto &entry : label_counts) {
        if (entry.second > max_votes) {
            max_votes = entry.second;
            predicted_label = entry.first;
        }
    }
 
    return predicted_label;
}

train()

void gpmp::ml::KNN::train	(	const std::vector< std::vector< double >> &	training_data,
		const std::vector< int > &	labels
	)

Trains the KNN model with the given training data and labels.

Parameters

training_data	The training data, a vector of feature vectors
labels	The corresponding labels for the training data

Definition at line 45 of file knn.cpp.

                                   {
    if (_training_data.size() != _labels.size() || _training_data.empty()) {
        throw std::invalid_argument("Invalid training data or labels.");
    }
 
    this->training_data = _training_data;
    this->labels = _labels;
}

Member Data Documentation

labels

std::vector<int> gpmp::ml::KNN::labels

private

Definition at line 84 of file knn.hpp.

training_data

std::vector<std::vector<double> > gpmp::ml::KNN::training_data

private

< The training data stored as feature vectors The corresponding labels for the training data

Definition at line 82 of file knn.hpp.

---

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

• include/openGPMP/ml/knn.hpp
• modules/ml/knn.cpp