Good question! One-hot encoding using pd.get_dummies(df) only one-hot encodes the categorical variables. The numerical variables are not affected.