Course contentsShow
← Data Science
Lesson 536 of 2,14514. Simple Linear RegressionFree lesson

Residual Standard Error (RSE)

Calculate RSE as an estimate of residual variance; understand it as average prediction error magnitude.

Residual Standard Error (RSE)

You'll learn: How to quantify the typical size of your prediction errors using Residual Standard Error.

What RSE Tells You

After fitting a simple linear regression line, you have residuals—the vertical distances between each actual data point and your predicted value on the line. But how wrong is your model on average? That's where Residual Standard Error (RSE) comes in.

RSE measures the typical magnitude of your prediction errors. Think of it as the "standard deviation of the residuals." Just as standard deviation tells you how spread out data points are from the mean, RSE tells you how spread out data points are from your regression line.

The Formula

RSE is calculated as:

RSE = √(RSS / (n - 2))

Where:

  • RSS = Residual Sum of Squares (the sum of all squared residuals)
  • n = number of observations
  • n - 2 = degrees of freedom (we lose 2 because we estimated two parameters: slope and intercept)

Why It Matters

RSE is expressed in the same units as your response variable. If you're predicting house prices in dollars and your RSE is 15,000,thatmeansyourpredictionsaretypicallyoffbyabout15,000, that means your predictions are typically off by about 15,000.

Smaller RSE = Better predictions. A model with RSE of 5,000makesmoreaccuratepredictionsthanonewithRSEof5,000 makes more accurate predictions than one with RSE of 20,000.

RSE also estimates the variance of the irreducible error—the natural noise in your data that even a perfect model couldn't eliminate.

Key Takeaway: RSE quantifies your average prediction error in original units, giving you an intuitive sense of how accurate your regression model is in practice.