BetaRegression<T>

Models & Types · AiDotNet.Regression

Beta Regression for modeling proportions and rates bounded in (0, 1).

Inheritance: AsyncDecisionTreeRegressionBase<T> → BetaRegression<T>

For Beginners

When you need to predict proportions (like percentages), regular regression can give impossible results (negative values or values > 1). Beta Regression fixes this by:

  1. Always producing valid predictions between 0 and 1
  2. Naturally handling skewed proportions
  3. Allowing varying uncertainty (some predictions more reliable than others)

Example applications:

  • Predicting market share (e.g., “37% market share”)
  • Modeling test pass rates
  • Estimating probability scores
  • Analyzing biological concentrations

The model uses a “link function” to transform proportions to a scale where linear modeling works, then transforms predictions back to valid proportions.

How It Works

Beta Regression is the appropriate choice when your response variable is a continuous proportion or rate that must fall strictly between 0 and 1. It uses the Beta distribution and can model both the mean and precision as functions of covariates.

Reference: Ferrari, S.L.P., Cribari-Neto, F. (2004). “Beta Regression for Modelling Rates and Proportions”. Journal of Applied Statistics, 31(7), 799-815.

Example

using AiDotNet;
using AiDotNet.Data.Loaders;
using AiDotNet.Regression;
using AiDotNet.Tensors.LinearAlgebra;

double[][] features =
{
    new[] { 1.0, 2.0 }, new[] { 2.0, 3.0 }, new[] { 3.0, 4.0 },
    new[] { 4.0, 5.0 }, new[] { 5.0, 6.0 }, new[] { 6.0, 7.0 }
};
double[] targets = { 3.0, 5.0, 7.0, 9.0, 11.0, 13.0 };

var result = await new AiModelBuilder<double, Matrix<double>, Vector<double>>()
    .ConfigureModel(new BetaRegression<double>())
    .ConfigureDataLoader(DataLoaders.FromArrays(features, targets))
    .BuildAsync();

Console.WriteLine("Trained BetaRegression.");

Constructors

Constructor Summary
BetaRegression(BetaRegressionOptions options, IRegularization<T, Matrix<T>, Vector<T>> regularization) Initializes a new instance of BetaRegression.

Properties

Property Summary
T MeanIntercept { get; } Gets the mean model intercept.
T Precision { get; } Gets the precision (or its intercept if constant).
Vector<T> MeanCoefficients { get; } Gets the mean model coefficients.
Vector<T> PrecisionCoefficients { get; } Gets the precision model coefficients (if variable precision is enabled).
int NumberOfTrees { get; }

Methods

Method Summary
ModelMetadata<T> GetModelMetadata()
Task TrainAsync(Matrix<T> x, Vector<T> y)
Task<IParametricDistribution<T>[]> PredictDistributionsAsync(Matrix<T> input) Predicts full Beta distributions for each input sample.
Task<ValueTuple<Vector<T>, Vector<T>>> PredictIntervalAsync(Matrix<T> input, double confidenceLevel) Gets prediction intervals for each input sample.
Task<Vector<T>> PredictAsync(Matrix<T> input)