One-hot encode a categorical column for modelling

Question

Before modelling, the `plan` column has to become numeric dummy columns.

Write `encode(df)` that returns the DataFrame with `plan` replaced by one-hot columns named `plan_<value>`, dropping the first category to avoid collinearity, with the remaining columns as integers (0/1) and the original colum

Accepted Answer

Step 1: `pd.get_dummies(..., columns=['plan'])` encodes only the column you name and leaves `user_id` and `seats` untouched. Step 2: `drop_first=True` removes one category so the dummy columns are not perfectly collinear — important for linear models, unnecessary for trees. Step 3: `dtype=int` gives 0/1 integers; the pandas default is boolean, which some estimators and SQL loaders reject. Step 4: Categories are ordered alphabetically, so 'basic' is the one dropped and becomes the reference level.

One-hot encode a categorical column for modelling

More Pandas — Cleaning questions