|
Internet
|
You can link to a prediction applet that demonstrates the difficulty of identifying, without the aid of mathematics, the one least squares prediction line for a given scatterplot. (Prediction line is referred to here as the regression line.) The least squares prediction line minimizes the total of all squared predictive errors (or all squared deviations of points from the line) and therefore also -- just one computational step away -- the mean of all squared errors or the MSE referred to in this applet. Your task is to draw a prediction line whose MSE approaches the specified minimum MSE for the actual least squares prediction line obtained from the formula discussed in Section 10.4 of the text.
|