black_friday_challenge_pipeline

black_friday_challenge_pipeline.py

from sklearn.ensemble import RandomForestRegressor
from train.models.custom_transformers import BlackFridayPreprocess, BlackFridayIdTransformer, BlackFridayLabelEncoder
from sklearn.pipeline import Pipeline


def random_forest():
    pipeline = Pipeline(
        [
            ('preprocess', BlackFridayPreprocess()),
            ('id_filter', BlackFridayIdTransformer(
                min_freq={'Product_ID': 200, 'User_ID': 100})),
            ('label_encoder', BlackFridayLabelEncoder()),
            ('regressor', RandomForestRegressor(n_estimators=50, max_depth=15))
        ]
    )

    return pipeline