person
Rookie's submission
Model: runs ? home_runs + walks
Data range: HR 162–245, walks 480–560
Prediction for HR = 300, walks = 700:
ŷ = 956 runs
"A team this good would score 956 runs!"
1 · Why distrust this prediction?
A model only describes the region it was fit on. Push both predictors past every observed team and the linear pattern can simply break down.
2 · What's safer?
Interpolation (inside the data) is reliable; extrapolating on several predictors at once is even riskier than on one.
open_in_full
Graded — stay inside the cloud.
With multiple predictors, “inside the data? means inside the joint cloud of observed combinations — not just each variable's range.
