For this lab, we will be using the dataset in the Customer Analysis Business Case of the previous lab. This dataset can be found in files_for_lab folder. In this lab we will explore categorical data.
As in this lab we will keep working of the same dataset of the previous lab, please make a copy of the final Jupyter notebook of the previous lab in the current lab folder. Next, use Markdown to add a new section in the Jupyter notebook named Lab Comparing Regression Models. Then restart the Kernel and run all the previous cells. Finally, keep working of the same notebook according to the next instructions.
- Define X and y
- Import sklearn train_test_split and separate the data. Set test_size=0.30 and random_state=31
- Separate X_train and X_test into numerical and categorical (X_train_cat , X_train_num , X_test_cat , X_test_num)
- Encode the categorical variables X_train_cat and X_test_cat using the OneHotEncoder setup in the previous lab. Remember to use
.toarray()after.transform()to endup with a numpy array. Next, cast the resulting numpy arrays into pandas DataFrames. Make sure that the column names of the new dataframes are correctly setup usingencoder.get_feature_names_out()and the same indexes of X_train_cat and X_test_cat - Use X_train_num to fit a power transformer. Transform BOTH X_train_num and X_test_num. Next, cast the resulting numpy arrays as pandas dataframes. Make sure to set the correct columns names and to use the same indexes of X_train_num and X_test_num. Name the final resulting dataframes as: X_train_num_transformed_df and X_test_num_transformed_df
- Concat X_train_num_transformed_df and X_train_cat_encoded_df into X_train_new and X_test_num_transformed_df and X_test_cat_encoded_df into X_test_new
- Fit a MinMax scaler using X_train_new and transform X_train_new and X_test_new. Create new pandas dataframes from the resulting numpy arrays. Remember to set the correct columns names and indexes. Name the resulting dataframes as: X_train_new_scaled_df and X_test_new_scaled_df
- Train a simple linear regression model using X_train_new_scaled_df, and get the predictions for the train and test sets
- Create a function that given a model prediction and real values returns a pandas dataframe with the following table:
| Error_metric | Value |
|---|---|
| MAE | value |
| MSE | value |
| RMSE | value |
| MAPE | value |
| R2 | value |
- Evaluate the linear model predictions using the previous function on the TRAIN and TEST sets
- Now define a function that takes as an input: list of models, X_train and y_train to train several model (with default values) so we can train a lot of them without repeating code. The function must return the list of trained models.
- Use the function to train the following models (with default settings):
- Evaluate the models with the function created earlier in the TRAIN and TEST sets. Which model performs best with the default options?