Description
In this exercise, we predict the sale price of a house using its other characteristics. The
training data are in the file “housing train.csv”, and the test data are in “housing test.csv”.
The response is in the column “Sale price”. Among the 25 feature variables, some are
numeric features, such as living area square feet or first floor square feet, and some are
categorical features, such as the overall material and finish of the house or kitchen quality.
A detailed description of the variables is in “dictionary.txt”.
(a) Fit a linear model using least squares on the training data. Is there any potential
disadvantage of this model?
(b) Fit a lasso model on the training data and report the test error. When the 1SE rule
is applied, how many predictors are included in the model?
(c) Fit an elastic net model on the training data. Report the selected tuning parameters
and the test error.
(d) Fit a partial least squares model on the training data and report the test error. How
many components are included in your model?
(e) Which model will you choose for predicting the response? Why?
1 P8106 Homework 1
