RSquaredFitnessCalculator<T, TInput, TOutput>

Models & Types · AiDotNet.FitnessCalculators

A fitness calculator that uses R-Squared (R²) to evaluate model performance.

Inheritance: FitnessCalculatorBase<T, TInput, TOutput> → RSquaredFitnessCalculator<T, TInput, TOutput>

For Beginners

This calculator helps evaluate how well your model is performing by measuring the proportion of variance in your target variable that is explained by your model.

R-Squared (R²), also called the coefficient of determination, is:

  • A measure of how well your model explains the variation in your data
  • Expressed as a value typically between 0 and 1 (or 0% to 100%)
  • A higher value means your model explains more of the variation in your data

How R-Squared works:

  • R² = 1 means your model perfectly explains all the variation in your data
  • R² = 0 means your model doesn’t explain any of the variation (it’s no better than just predicting the average)
  • R² can sometimes be negative if your model performs worse than just predicting the average

Think of it like this: Imagine you’re predicting house prices:

  • If R² = 0.7, it means 70% of the variation in house prices is explained by your model
  • The remaining 30% is due to factors your model doesn’t capture

A simple way to understand R²:

  • If you always predicted the average house price, you’d have an R² of 0
  • If you could predict every house price exactly right, you’d have an R² of 1
  • Your model’s R² tells you how much better it is than just predicting the average

Key characteristics of R²:

  • Higher values are better (1 would be perfect)
  • It’s scale-independent (it doesn’t matter if you’re predicting dollars or millions of dollars)
  • It helps you understand how much of the variation your model captures
  • It can be misleading if you have a small sample size or too many features

When to use R²:

  • When you want to know how much of the variation your model explains
  • When you want a metric that’s easy to interpret (0% to 100% explained)
  • When comparing different models for the same problem
  • For regression problems (predicting continuous values)

R² is one of the most popular metrics for regression tasks because it provides an intuitive measure of how well your model captures the patterns in your data.

Constructors

Constructor Summary
RSquaredFitnessCalculator(DataSetType dataSetType) Initializes a new instance of the RSquaredFitnessCalculator class.