R Programming By Example

Checking homoscedasticity with residual plots

Homoscedasticity simply means that we need the data to have constant variance in our residuals. To check for it, we can use the plot(fit) function call. However, this will show one plot at a time asking you to hit Enter on your keyboard to show the next one. This kind of mechanism is not friendly to the automation processes we are creating. So we need a little adjustment. We will use the par(mfrow = c(2, 2)) call to tell the plot() function to graph all four plots at the same time and show it in a single image. We wrap the command around our already familiar mechanism to save PNGs around the fit_plot() function, and we're all set:

fit_plot <- function(fit, save_to = "") { if (not_empty(save_to)) ...