preprocess.yaml

drop_columns :
  - "energy_performance_category"
  - "ghg_category"
encoding :
  frequency_encoding :
    - "city"
  quantile_encoding : null
  label_encoding : null
  target_encoding : null
  hot_encoding: True
imputation :
  constant_imputation :
    constant_floor: False
    constant_land_size: False
    constant_energy_performance_value: False
    constant_ghg_value: False
    constant_bedrooms: False
    constant_exposition: False
  iter_imputation: False
  mean_imputation: False
  mini_imputation: False
polar :
  add_polar_coordinates: True
  add_polar_rotation:
    - 10
    - 30
    - 45
    - 60
    - 70
    - 90
    - 145
    - 180
  add_polar_pca: True
  add_exposition_orientation: False
geo:
  add_geopopulation: False
  add_geopopulation_2: True # good
  add_distance_to_city_center: False
  add_geo: True
  geodata_list : #null
    - index
    - "num_train_station"
    - "num_gas_station"
    - "user_ratings_total"
images:
  nima :
    add_classification_quality: False
    classification_threshold: 0
    images_features: null
  caption :
    add_captions: True
    vectorize_all: False
    max_features: 0
    has_items:
      - "pool"
      - "luxury"
      # - "property image" # no
      # - "jacuzzi" # no
      # - "pharmacy" no
      # - "parking" no
      # - "million" no
      # - "balcony" no
      # - "garden" not this
      # - "ocean" # don't add
    count_items:
      # - "bathroom" don't add
      - "bedroom"
      - "living"
      - "villa"
      # - "sea"
      # - "large house" # don't add
      # - "balcony" # don't add
      # - "kitchen" don't add
      # - "dining" # don't add
feature_transformation: False
features_interactions: False
scaling :
  standard_scaling: False
  robust_scaling: True
  power_scaling: False
target_transformation: True
final_dump:
  - 'approximate_latitude'