etl_config.py

RAW_CONFIG = {
    'numeric': [
        "land_area",
        "building_area",
        "parking_space_area",
        "main_building_area",
        "balcony_area",
        "attached_building_area",
        "transfer_floor_level",
        "total_floors",
        "building_age"
        "NTDLRT_mrt_dist",
        "KRTC_mrt_dist",
        "TYMC_mrt_dist",
        "KLRT_mrt_dist",
        "TRTC_mrt_dist",
        "train_dist",
        "hospital_dist",
        "high_school_dist",
        'university_dist',
        'senior_high_school_dist',
    ],
    'target_mean': [],
    'target_count': ['city', 'area', 'area_road'],
    'target_var': [],
    "freq_cat": [
        "area_road",
        "number_of_parking_spaces",
        "usage_zone",
        "primary_use",
        "main_material",
        "building_type",
        "NTDLRT_mrt_name",
        "TYMC_mrt_name",
        "TRTC_mrt_name",
        'train_name',
        'hospital_name',
        "high_school_name",
        'university_name',
        'senior_high_school_name',
        'is_notes',
    ],
    'cat': [
        "city",
        "area",
        "area_road",
        "number_of_parking_spaces",
        "primary_use",
        "main_material",
        "building_type",
        "NTDLRT_mrt_name",
        "TYMC_mrt_name",
        "TRTC_mrt_name",
        'train_name',
        'hospital_name',
        "high_school_name",
        'university_name',
        'senior_high_school_name',
        "community_name",
        "recreation_center_name",
    ],
    'remove_cols': [
        'district', 'road', 'x_coordinate', 'y_coordinate', 'notes', 'id',
        'lon', 'lat', 'KLRT_mrt_name', 'KRTC_mrt_name', 'is_notes',
        'usage_zone'
    ]
}