Calculate Coefficient of Determination R²

Enter Sum of Squares Regression (SSR) and Sum of Squares Error (SSE).

Explained sum of squares $\text{SSR}$ (e.g. 75).
Unexplained sum of squares $\text{SSE}$ (e.g. 25).

Calculation Results

Primary Metric Output --
Metric Breakdown 1--
Metric Breakdown 2--
Metric Breakdown 3--
Metric Breakdown 4--
Metric Breakdown 5--
Mathematical Standard--

Calculated using coefficient of determination equations: R^2 = \frac{\text{SSR}}{\text{SST}} = 1 - \frac{\text{SSE}}{\text{SST}}

*Note: R-squared measures the proportion of total variance in dependent variable Y explained by regression model inputs.

Quick Summary

The R-Squared Calculator evaluates the coefficient of determination ($R^2 = \frac{\text{SSR}}{\text{SST}}$), percentage explained variance, and unexplained residual error percentage.

Formula Explanation

R^2 = \frac{\text{SSR}}{\text{SST}} = 1 - \frac{\text{SSE}}{\text{SST}}
\text{SST} = \text{SSR} + \text{SSE}

How It Works

R-squared ($R^2$) measures goodness-of-fit in regression analysis. The R-Squared Calculator divides explained Sum of Squares Regression (SSR) by Total Sum of Squares (SST = SSR + SSE).

Step-by-Step Worked Example

Practical Problem: A statistical model has $\text{SSR} = 75$ and $\text{SSE} = 25$. Calculate $R^2$ coefficient of determination.

  1. Step 1: Calculate Total Sum of Squares (SST): $\text{SST} = \text{SSR} + \text{SSE} = 75 + 25 = \mathbf{100.0}$.
  2. Step 2: Divide SSR by SST: $R^2 = 75 / 100 = \mathbf{0.7500}$.
  3. Step 3: Convert to percentage: $0.7500 \times 100 = \mathbf{75.00\%}$.
  4. Step 4: Calculate unexplained variance percentage: $100\% - 75\% = \mathbf{25.00\%}$.
  5. Step 5: Interpretation: The model explains 75.00% of the total variance in the outcome variable $Y$.

Real-World Calculation Examples

Scenario 1: Model SSR = 75, SSE = 25

Parameters: SSR = 75, SSE = 25
Result: R² = 0.7500 (75.00% variance explained).

Scenario 2: High Precision Physics Experiment

Parameters: SSR = 99, SSE = 1
Result: R² = 0.9900 (99% precision fit).

Scenario 3: Social Science Survey Behavior

Parameters: SSR = 20, SSE = 80
Result: R² = 0.2000 (20% variance explained).

Scenario 4: Stock Market Beta CAPM Model

Parameters: Market variance fit
Result: R-squared benchmark evaluation.

Key Benefits of Using This Calculator

Goodness-of-Fit Quantification

Measures model explanatory power on a standardized 0% to 100% scale.

Explained vs Unexplained Breakdown

Outputs both explained model percentage and residual error percentage.

Sum of Squares Partitioning

Displays Total Sum of Squares (SST = SSR + SSE).

100% Free & Client-Side

Executes locally in your browser with zero latency or web server transmission.

Frequently Asked Questions (FAQ)

What is R-squared ($R^2$)?

R-squared ($R^2$) is the coefficient of determination measuring the proportion of variance in dependent outcome $Y$ explained by regression inputs.

What is range of R-squared?

In OLS linear regression, $R^2$ ranges strictly from $0.0$ (0%) to $1.0$ (100%).

Can R-squared be negative?

In standard OLS without forced zero-intercept, $R^2 \ge 0$. In non-linear regression or forced zero-intercept models, $R^2$ can technically be negative.

How do I calculate R-squared in Excel?

Use formula =RSQ(known_y, known_x).

What is a "good" R-squared value?

In physics, $R^2 > 0.90$ is common; in social sciences and economics, $R^2 \ge 0.30$ or $0.50$ is often considered strong.

What is difference between R-squared and Pearson $r$?

In simple linear regression, $R^2 = r^2$ (the square of Pearson correlation coefficient $r$).

Does a high R-squared prove causation?

No — a high $R^2$ indicates high predictive accuracy, but does NOT prove causal direction.

What is Total Sum of Squares (SST)?

$\text{SST} = \sum (y_i - \bar{y})^2$, representing total variation in outcome $Y$ around its mean.

What is Residual Sum of Squares (SSE)?

$\text{SSE} = \sum (y_i - \hat{y}_i)^2$, representing unexplained model residual error.

Why is Adjusted R-squared preferred for multiple regression?

Adjusted $R^2$ adjusts for number of predictors $k$, preventing artificial inflation when adding extra variables.