07_Modelling_Classic

Question 1

Front

Answer 1

Back

Question 2

What does lm() do and how do you use it?

Answer 2

Fits linear models via OLS for regression and inference.

Code:
fit <- lm(mpg ~ wt + hp, data = mtcars)
summary(fit); confint(fit)
predict(fit, newdata = data.frame(wt=3, hp=110), interval=’prediction’)

Notes:
Check assumptions with residual diagnostics (plot(fit), qqnorm/qqline).

Question 3

How do you fit a logistic regression with glm()?

Answer 3

Use glm(family = binomial) for binary outcomes.

Code:
fit <- glm(am ~ mpg + wt, data = mtcars, family = binomial())
summary(fit)
predict(fit, newdata = mtcars, type = ‘response’) # probabilities

Notes:
Interpret coefficients on log-odds scale; use broom::tidy() for tidy output.

Question 4

How do you compare nested models? (anova, AIC/BIC)

Answer 4

Use anova(m1, m2) and AIC/BIC for information criteria.

Code:
m1 <- lm(mpg ~ wt, data=mtcars)
m2 <- lm(mpg ~ wt + hp, data=mtcars)
anova(m1, m2); AIC(m1, m2); BIC(m1, m2)

Question 5

How do you extract fitted values and residuals?

Answer 5

Use fitted() and residuals().

Code:
fit <- lm(mpg ~ wt, data = mtcars)
head(fitted(fit)); head(residuals(fit))