#model-misspecification

Question 1: Single choice

mediumAI-checkedLinear Models & Regularization

A large dataset has inputs in $[0, 1]$ , and you want the least-squares line that best fits the whole dataset, without assuming the relation is linear. You can either request random examples from the dataset or ask for the output at an input of your choice. Which query strategy should you use?

Query only the extreme inputs $x = 0$ and $x = 1$ 1
Query inputs evenly spaced over $[0, 1]$ 2
Draw random examples from the dataset3
Query the inputs where the current line has the largest predictive variance4

#model-misspecification

Timed test