NGBoostRegression<T>

Models & Types · AiDotNet.Regression

NGBoost (Natural Gradient Boosting) for probabilistic regression.

Inheritance: AsyncDecisionTreeRegressionBase<T> → NGBoostRegression<T>

For Beginners

Traditional regression gives you a point prediction like “the house price is $300,000.” But NGBoost tells you “the house price follows a normal distribution with mean $300,000 and standard deviation $50,000.”

This uncertainty information is valuable because it tells you how confident the model is. A prediction with small uncertainty means the model is confident. A prediction with large uncertainty means you should be more cautious.

Key benefits:

  • Quantifies prediction uncertainty
  • Can use different distributions for different types of data
  • Uses natural gradients for stable, efficient learning

How It Works

NGBoost is an algorithm for probabilistic prediction that uses natural gradients to efficiently and directly optimize a proper scoring rule. Instead of predicting a single value, NGBoost predicts a full probability distribution.

Reference: Duan, T., et al. “NGBoost: Natural Gradient Boosting for Probabilistic Prediction” (2019). https://arxiv.org/abs/1910.03225

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 NGBoostRegression<double>())
    .ConfigureDataLoader(DataLoaders.FromArrays(features, targets))
    .BuildAsync();

Console.WriteLine("Trained NGBoostRegression.");

Constructors

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

Properties

Property Summary
NGBoostDistributionType DistributionType { get; } Gets the distribution type used by this model.
int NumberOfTrees { get; }

Methods

Method Summary
ModelMetadata<T> GetModelMetadata()
Task TrainAsync(Matrix<T> x, Vector<T> y) Trains the NGBoost model using natural gradient boosting.
Task<IParametricDistribution<T>[]> PredictDistributionsAsync(Matrix<T> input) Predicts full probability distributions for each input sample.
Task<Matrix<T>> PredictQuantilesAsync(Matrix<T> input, Vector<double> quantiles) Predicts quantiles 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) Predicts the mean of the distribution for each input sample.