Prediction and model comparison in R

Using a fitted model and testing it against a simpler one.

model <- lm(mpg ~ wt + hp, data = mtcars)

new_cars <- data.frame(wt = c(2.5, 3.5), hp = c(110, 180))
point <- predict(model, newdata = new_cars)
with_interval <- predict(model, new_cars, interval = "confidence")

rmse <- sqrt(mean(resid(model)^2))
aic <- AIC(model)
reduced <- update(model, . ~ . - hp)
comparison <- anova(reduced, model)

How it works

1. predict with newdata scores unseen rows.
2. interval = "confidence" adds uncertainty bounds.
3. anova compares nested models; AIC scores fit.

Keywords and builtins used here

• AIC
• anova
• c
• lm
• mean
• predict
• resid
• sqrt
• update

The run, in numbers

Lines

10

Characters to type

339

Tokens

99

Three-star pace

100 tpm

At the three-star pace of 100 tokens a minute, this run takes about 59 seconds.

Type this snippet

Step 4 of 4 in Correlation & regression, step 9 of 11 in Statistics.

← Previous Next →