get_mse: Extract Mean Squared Error (MSE) from Fitted Regression Model

In dvmisc: Faster Computation of Common Statistics and Miscellaneous Functions

Description

The MSE, defined as the sum of the squared residuals divided by n-p (n = number of observations, p = number of regression coefficients), is an unbiased estimator for the error variance in a linear regression model. This is a convenience function that extracts the MSE from a fitted lm or glm object. The code is rev(anova(model.fit)$"Mean Sq")[1] if model.fit is a lm object and sum(model.fit$residuals^2) / model.fit$df.residual if model.fit is a glm object.

Usage

get_mse(model.fit, var.estimate = FALSE)

Arguments

model.fit	Fitted regression model returned from lm or glm.
var.estimate	If TRUE, function returns a variance estimate for the error variance, defined as 2 * MSE^2 / (n - p).

Value

If var.estimate = FALSE, numeric value indicating the MSE; if var.estimate = TRUE, named numeric vector indicating both the MSE and a variance estimate for the error variance.

Examples

# Generate 100 values: Y = 0.5 + 1.25 X + e, e ~ N(0, 1)
set.seed(123)
x <- rnorm(100)
y <- 0.5 + 1.25 * x + rnorm(100, sd = 1)

# Fit regression model using lm and using glm
lm.fit <- lm(y ~ x)
glm.fit <- glm(y ~ x)

# Extract MSE from lm.fit and glm.fit
get_mse(lm.fit)
get_mse(glm.fit)

dvmisc documentation built on May 2, 2019, 5:51 p.m.