Multivariate Linear Regression

Multivariate linear regression for Dart with support for multiple outputs and optional intercept, implemented using Golub-Reinsch Singular Value Decomposition.

Inspired by ml-matrix and regression-multivariate-linear Node.js libraries.

Installation

In order to start using Multivariate Linear Regression you must have the Dart SDK installed on your machine.

Install via dart pub add:

dart pub add multivariate_linear_regression

Usage

import 'package:multivariate_linear_regression/multivariate_linear_regression.dart';

void main() {
  final x = [
    [0.0, 0.0],
    [1.0, 2.0],
    [2.0, 3.0],
    [3.0, 4.0],
  ];

  final y = [
    [0.0, 0.0, 0.0],
    [2.0, 4.0, 3.0],
    [4.0, 6.0, 5.0],
    [6.0, 8.0, 7.0],
  ];

  final mlr = MultivariateLinearRegression(
    x: x,
    y: y,
  );

  print(mlr.predict([3.0, 3.0]));
  print(mlr.predictBatch([[1.0, 2.0], [2.0, 3.0]]));
  print(mlr.weights);
  print(mlr.stdError);
  print(mlr.stdErrors);
  print(mlr.tStats);
  print(mlr.toJson());
}

API Overview

Constructor

MultivariateLinearRegression({
  required List<List<double>> x,
  required List<List<double>> y,
  bool intercept = true,
  bool statistics = true,
})

Creates a multivariate linear regression model.

x - Input feature matrix (rows = samples, columns = features)
y — Output matrix (rows = samples, columns = targets)
intercept — Includes a bias (intercept) term when set to true
statistics — Enables computation of additional metrics (standard errors, t-stats, etc.)

Load Existing Model

factory MultivariateLinearRegression.load(MultivariateLinearRegression model)

Reconstructs a trained model from previously trained model

Prediction

List<double> predict(List<double> input)

Returns predicted outputs for a single input vector.

List<List<double>> predictBatch(List<List<double>> inputs)

Returns predictions for multiple input rows.

Coefficients & Metrics

List<List<double>> get weights

Matrix of regression coefficients (includes intercept if enabled).

double get stdError

Overall standard error of the model.

List<List<double>> get stdErrors

Standard error for each coefficient.

List<List<double>> get tStats

T-statistics corresponding to each coefficient.

List<List<double>> get stdErrorMatrix

Covariance matrix of the coefficients.

Available only when statistics = true

Serialization

Map<String, dynamic> toJson()

Serializes the model into a JSON-compatible format, including statistics when enabled.

Continuous Integration

Multivariate Linear Regression comes with a built-in GitHub Actions workflow powered by Very Good Workflows.

On each pull request and push, the CI formats, lints, and tests the code. The project uses Very Good Analysis for a strict set of analysis rules. Code coverage is enforced using Very Good Coverage.

Running Tests

To run all unit tests and generate coverage:

dart pub global activate coverage 1.15.0
dart test --coverage=coverage
dart pub global run coverage:format_coverage --lcov --in=coverage --out=coverage/lcov.info

To view the coverage report using lcov:

genhtml coverage/lcov.info -o coverage/
open coverage/index.html

Support

If you find this package useful, please consider supporting it:

Like the package on pub.dev
Star the GitHub repository

Your support helps improve the project and keeps it actively maintained 😊