Home » Coefficient of Determination (R²)

Coefficient of Determination (R²)

1900
  • Karl Pearson
Statistician analyzing regression model data in an office setting.

A statistic indicating the goodness of fit of a model, representing the proportion of the variance in the dependent variable that is predictable from the independent variable(s). An R² of 1 indicates a perfect fit, while 0 indicates no linear relationship. It is calculated as \(R^2 \equiv 1 – \frac{SS_{res}}{SS_{tot}}\), where \(SS_{res}\) is the residual sum of squares.

The coefficient of determination, R-squared, is a key metric for evaluating regression models. It provides an intuitive measure of how much of the variability in the outcome is captured by the model. It is derived from two key components. The first is the Total Sum of Squares (\(SS_{tot} = \sum_i (y_i – \bar{y})^2\)), which measures the total variance in the dependent variable \(y\). The second is the Residual Sum of Squares (\(SS_{res} = \sum_i (y_i – \hat{y}_i)^2\)), which measures the variance left unexplained by the model, where \(\hat{y}_i\) is the predicted value.

The formula \(R^2 = 1 – SS_{res}/SS_{tot}\) can be interpreted as the percentage of total variance that is ‘explained’ by the regression model. For instance, an R² of 0.75 means that 75% of the variability in the outcome can be accounted for by the predictors in the model. In simple linear regression, R² is simply the square of Pearson’s correlation coefficient (r) between the observed and predicted values.

However, R² has a significant limitation: it never decreases when a new predictor variable is added to the model, even if the new variable is irrelevant. This can be misleading and encourage overfitting. To counteract this, the Adjusted R-squared is often used. It modifies the R² value to account for the number of predictors in the model, providing a more accurate measure of goodness of fit for multiple regression.

UNESCO Nomenclature: 1209
– Statistics

Type

Abstract System

Disruption

Substantial

Usage

Widespread Use

Precursors

  • Concept of variance and standard deviation
  • Method of least squares
  • Pearson’s product-moment correlation coefficient
  • Analysis of variance (ANOVA) principles

Applications

  • evaluating the performance of predictive models in science and engineering
  • model selection in econometrics and social sciences
  • quantifying the proportion of variance explained by a set of predictors
  • validating financial models for risk assessment

Patents:

NA

Potential Innovations Ideas

Professionals (100% free) Membership Required

You must be a Professionals (100% free) member to access this content.

Join Now

Already a member? Log in here
Related to: r-squared, coefficient of determination, goodness of fit, model evaluation, explained variance, sum of squares, regression diagnostics, statistical significance, adjusted r-squared, correlation.

Leave a Reply

Your email address will not be published. Required fields are marked *

AVAILABLE FOR NEW CHALLENGES
Mechanical Engineer, Project, Process Engineering or R&D Manager
Effective product development

Available for a new challenge on short notice.
Contact me on LinkedIn
Plastic metal electronics integration, Design-to-cost, GMP, Ergonomics, Medium to high-volume devices & consumables, Lean Manufacturing, Regulated industries, CE & FDA, CAD, Solidworks, Lean Sigma Black Belt, medical ISO 13485

We are looking for a new sponsor

 

Your company or institution is into technique, science or research ?
> send us a message <

Receive all new articles
Free, no spam, email not distributed nor resold

or you can get your full membership -for free- to access all restricted content >here<

Historical Context

(if date is unknown or not relevant, e.g. "fluid mechanics", a rounded estimation of its notable emergence is provided)

Related Invention, Innovation & Technical Principles

Scroll to Top

You May Also Like