Idea: another example of interactions for exercise 1.e

[necantis]

Weight has been decreasing over time in some zones.
lm.fit_complex <- lm(mpg ~ weightyearorigin, data=Auto)

Since the number of parameters is fairly high, we should check the Adjusted R2 is 0.83. All parameters are statistically significant, although some effects are fairly small
summary(lm.fit_complex)

[kbenoit]

Do you mean this?

lm(mpg ~ weight * year * origin, data = Auto)

[necantis]

Yes. I copied the R code on github and it lost the * .