params.yaml

random_seed: 230213
version: '0.2.0'
title: 'Apartment price model'
description: 'A pricing model, which can predict the acceptable per night price for Airbnb apartment based on its properties and the offered amenities'

data:
  source_url: 'http://data.insideairbnb.com/spain/catalonia/barcelona/2022-12-11/data/listings.csv.gz'
  raw_data_path: 'data/raw'
  interim_data_path: 'data/interim'
  processed_data_path: 'data/processed'
  train_data_file: 'train.csv'
  test_data_file: 'test.csv'
  categorical_feature_names_file: 'categorical_feature_names.csv'
  features:
    - host_is_superhost
    - neighbourhood_group_cleansed
    - property_type
    - room_type
    - accommodates
    - bathrooms_text
    - bedrooms
    - beds
    - number_of_reviews
  target: 'price'
  test_split_ratio: 0.2

data_cleaning:
  feature_limits:
    accommodates: 11
    beds: 11
    bedrooms: 6
  target_limit: 500

model:
  categorical_features:
    - host_is_superhost
    - neighbourhood_group_cleansed
    - property_type
    - room_type
    - accommodates
    - bathrooms_text
    - bedrooms
    - beds

  path: 'models'
  report_path: 'reports'
  eval_hist_file: 'lgbm_regressor_eval.csv'
  model_file: 'lgbm_regressor.txt'
  column_transformer_file: 'column_transformer.pkl'
  model_performance_file: 'lgbm_regressor_performance.csv'