SimpleLinearRegression<T>

Models & Types · AiDotNet.Prototypes

Simple linear regression model for prototype validation. Demonstrates GPU acceleration for traditional machine learning algorithms.

How It Works

Linear regression model: y = X @ weights + bias

This implementation uses vectorized operations throughout, enabling GPU acceleration when using float type with GPU engine enabled.

Constructors

Constructor Summary
SimpleLinearRegression(int numFeatures) Initializes a new instance of the SimpleLinearRegression.

Properties

Property Summary
bool IsTrained { get; } Gets whether the model has been trained.
int NumFeatures { get; } Gets the number of features.

Methods

Method Summary
PrototypeVector<T> GetWeights() Gets the learned weights.
PrototypeVector<T> PredictBatch(PrototypeVector<T> X, int numSamples) Makes predictions for a batch of samples.
T ComputeMSE(PrototypeVector<T> predictions, PrototypeVector<T> targets) Computes Mean Squared Error.
T ComputeR2Score(PrototypeVector<T> predictions, PrototypeVector<T> targets) Computes R² score (coefficient of determination).
T GetBias() Gets the learned bias.
T Predict(PrototypeVector<T> features) Makes a prediction for a single sample.
string ToString() Returns a string representation of the model.
void Train(PrototypeVector<T> X, PrototypeVector<T> y, int numSamples, double learningRate, int numEpochs, bool verbose) Trains the model using gradient descent.