File size: 16,088 Bytes
bcd17c2 5089ff4 933cc13 bcd17c2 5089ff4 6267f13 5089ff4 6267f13 5089ff4 0120bb1 5089ff4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 |
import os
os.environ["PGEOCODE_CACHE_DIR"] = "/tmp/pgeocode"
import pandas as pd
import numpy as np
from sklearn.base import BaseEstimator, TransformerMixin
from scipy.stats import gaussian_kde
from sklearn.preprocessing import StandardScaler
from sklearn.neighbors import NearestNeighbors
from datetime import datetime
import pgeocode
# IMMEDIATELY after importing pgeocode, force its STORAGE_DIR
# This must be done before any pgeocode.Nominatim() calls
try:
pgeocode.STORAGE_DIR = "/tmp/pgeocode"
# Ensure the directory exists as pgeocode might not create it if overridden this way
os.makedirs(pgeocode.STORAGE_DIR, exist_ok=True)
print(f"DEBUG: Successfully forced pgeocode.STORAGE_DIR to '{pgeocode.STORAGE_DIR}' and ensured directory exists.")
except Exception as e:
print(f"ERROR: Failed to force pgeocode.STORAGE_DIR or create directory: {e}")
class DataCleaner(BaseEstimator, TransformerMixin):
def __init__(self):
self.col_types = {
'id': 'int', 'type': 'str', 'subtype': 'str', 'bedroomCount': 'int',
'bathroomCount': 'int', 'province': 'str', 'locality': 'str',
'postCode': 'int', 'habitableSurface': 'float', 'hasBasement': 'int',
'buildingCondition': 'str', 'buildingConstructionYear': 'int',
'hasLift': 'int', 'floodZoneType': 'str', 'heatingType': 'str',
'hasHeatPump': 'int', 'hasPhotovoltaicPanels': 'int',
'hasThermicPanels': 'int', 'kitchenType': 'str', 'landSurface': 'float',
'hasLivingRoom': 'int', 'livingRoomSurface': 'float', 'hasGarden': 'int',
'gardenSurface': 'float', 'parkingCountIndoor': 'int',
'parkingCountOutdoor': 'int', 'hasAirConditioning': 'int',
'hasArmoredDoor': 'int', 'hasVisiophone': 'int', 'hasOffice': 'int',
'toiletCount': 'int', 'hasSwimmingPool': 'int', 'hasFireplace': 'int',
'hasTerrace': 'int', 'terraceSurface': 'float', 'terraceOrientation': 'str',
'epcScore': 'str', 'facadeCount': 'int'
}
self.kitchenType_mode = None
def fit(self, X, y=None):
#self.kitchenType_mode = X['kitchenType'].mode()[0]
return self
def transform(self, X):
df = X.copy()
# Drop unnecessary columns
df = df.drop(columns=[col for col in ["Unnamed: 0", "url"] if col in df.columns])
df = df.drop(columns=[col for col in ['monthlyCost', 'hasBalcony', 'accessibleDisabledPeople',
'roomCount', 'diningRoomSurface', 'streetFacadeWidth',
'gardenOrientation', 'kitchenSurface', 'floorCount',
'hasDiningRoom', 'hasDressingRoom'] if col in df.columns])
# Handle binary columns
binary_cols = [
'hasBasement', 'hasLift', 'hasHeatPump', 'hasPhotovoltaicPanels',
'hasAirConditioning', 'hasArmoredDoor', 'hasVisiophone', 'hasOffice',
'hasSwimmingPool', 'hasFireplace', 'parkingCountIndoor', 'parkingCountOutdoor',
'hasAttic', 'hasThermicPanels'
]
for col in binary_cols:
df[col] = df[col].map({True: 1, False: 0, 'True': 1, 'False': 0, 'YES': 1, 'NO': 0}).fillna(0).astype(int)
# Handle dependent columns
df['hasLivingRoom'] = df['hasLivingRoom'].map({True: 1, False: 0, 'True': 1, 'False': 0, 'YES': 1, 'NO': 0})
df.loc[df['hasLivingRoom'].isna(), 'hasLivingRoom'] = df['livingRoomSurface'].notnull().astype(int)
df['hasGarden'] = df['hasGarden'].map({True: 1, False: 0, 'True': 1, 'False': 0, 'YES': 1, 'NO': 0})
df.loc[df['hasGarden'].isna(), 'hasGarden'] = df['gardenSurface'].notnull().astype(int)
df['hasTerrace'] = df['hasTerrace'].map({True: 1, False: 0, 'True': 1, 'False': 0, 'YES': 1, 'NO': 0})
df.loc[df['hasTerrace'].isna(), 'hasTerrace'] = df['terraceSurface'].notnull().astype(int)
# Set surfaces to 0 when feature is not present
df.loc[df['hasLivingRoom'] == 0, 'livingRoomSurface'] = 0
df.loc[df['hasGarden'] == 0, 'gardenSurface'] = 0
df.loc[df['hasTerrace'] == 0, 'terraceSurface'] = 0
df.loc[df['hasTerrace'] == 0, 'terraceOrientation'] = 0
# Handle facade count
df['facadeCount'] = df['facadeCount'].fillna(-1)
# Fill missing values
df['bedroomCount'] = df['bedroomCount'].fillna(-1).astype(float)
df['bathroomCount'] = df['bathroomCount'].fillna(-1).astype(float)
df['toiletCount'] = df['toiletCount'].fillna(-1).astype(float)
# Drop habitable surface na
df = df.dropna(subset=['habitableSurface'])
# Fill other missing values
df['buildingCondition'] = df['buildingCondition'].fillna('NOT_MENTIONED')
df['floodZoneType'] = df['floodZoneType'].fillna('NON_FLOOD_ZONE')
df['heatingType'] = df['heatingType'].fillna(df['heatingType'].mode()[0])
df['hasThermicPanels'] = df['hasThermicPanels'].fillna(0.0)
df['kitchenType'] = df['kitchenType'].fillna(df['kitchenType'].mode()[0])
df['landSurface'] = df['landSurface'].fillna(df['landSurface'].median())
df['livingRoomSurface'] = df['livingRoomSurface'].fillna(df['livingRoomSurface'].median())
# Transform building construction year into age and fillna(-1)
current_year = datetime.now().year
df['buildingAge'] = current_year - df['buildingConstructionYear']
df['buildingAge'] = df['buildingAge'].fillna(-1)
# Handle terrace surface and orientation
median_terrace = df.loc[(df['hasTerrace'] == 1) & (df['terraceSurface'].notnull()), 'terraceSurface'].median()
df.loc[(df['hasTerrace'] == 1) & (df['terraceSurface'].isna()), 'terraceSurface'] = -1
df.loc[(df['hasTerrace'] != 1) & (df['terraceSurface'].isna()), 'terraceSurface'] = 0
mode_terrace = df.loc[(df['hasTerrace'] == 1), 'terraceOrientation'].mode()[0]
df.loc[(df['hasTerrace'] == 1) & (df['terraceOrientation'].isna()), 'terraceOrientation'] = 'NOT_MENTIONED'
df.loc[(df['hasTerrace'] != 1) & (df['terraceOrientation'].isna()), 'terraceOrientation'] = 'NO_TERRACE'
# Convert data types
for col, dtype in self.col_types.items():
if col in df.columns:
if pd.api.types.is_integer_dtype(dtype):
df[col] = df[col].fillna(0).astype(dtype)
else:
df[col] = df[col].astype(dtype)
return df
class FeatureEngineer(BaseEstimator, TransformerMixin):
def __init__(self):
self.epc_mapping = {
'Flanders': {
'A++': 0, 'A+': 0, 'A': 100, 'B': 200, 'C': 300,
'D': 400, 'E': 500, 'F': 600, 'G': 700
},
'Wallonia': {
'A++': 0, 'A+': 50, 'A': 90, 'B': 170, 'C': 250,
'D': 330, 'E': 420, 'F': 510, 'G': 600
},
'Bruxelles': {
'A++': 0, 'A+': 0, 'A': 45, 'B': 95, 'C': 145,
'D': 210, 'E': 275, 'F': 345, 'G': 450
}
}
def fit(self, X, y=None):
return self
def transform(self, X):
df = X.copy()
if 'price' in df.columns:
# Filter out extremely high prices
high_price_count = (df['price'] > 1500000).sum()
df = df[df['price'] <= 1500000]
# Check for problematic values
zero_price = (df['price'] <= 0).sum()
zero_surface = (df['habitableSurface'] <= 0).sum()
# Handle problematic values
if zero_price > 0:
df.loc[df['price'] <= 0, 'price'] = np.nan
if zero_surface > 0:
df.loc[df['habitableSurface'] <= 0, 'habitableSurface'] = np.nan
# Add isHouse feature
df['isHouse'] = (df['type'] == 'HOUSE').astype(int)
# Add region information first
def get_region(zip_code):
if 1000 <= zip_code <= 1299:
return "Bruxelles"
elif 1300 <= zip_code <= 1499 or 4000 <= zip_code <= 7999:
return "Wallonia"
else:
return "Flanders"
df['region'] = df['postCode'].apply(get_region)
if 'price' in df.columns:
# Now add price per m2
df['pricePerM2'] = df['price'] / df['habitableSurface']
# Handle inf values
df['pricePerM2'] = df['pricePerM2'].replace([np.inf, -np.inf], np.nan)
# Fill NaN values with median by region
df['pricePerM2'] = df['pricePerM2'].fillna(-1)
# Convert EPC score
df['epcScore'] = df.apply(lambda row: self.epc_mapping.get(row['region'], {}).get(row['epcScore'], None), axis=1)
df['epcScore'] = df['epcScore'].fillna(-1)
# Convert building condition
condition_rating = {
'to restore': 0, 'to renovate': 1, 'to be done up': 2,
'good': 3, 'just renovated': 4, 'as new': 5
}
df['buildingCondition'] = (df['buildingCondition'].astype(str).str.strip().str.lower()
.map(condition_rating).fillna(-1).astype(int))
# Convert flood zone type
df['floodZoneType'] = (df['floodZoneType'] != 'NON_FLOOD_ZONE').astype(int)
return df
class CategoricalEncoder(BaseEstimator, TransformerMixin):
def __init__(self):
self.categorical_columns = ['province', 'heatingType', 'kitchenType', 'subtype', 'terraceOrientation']
def fit(self, X, y=None):
return self
def transform(self, X):
df = X.copy()
# One-hot encode categorical columns
for col in self.categorical_columns:
if col in df.columns:
df = pd.get_dummies(df, columns=[col], prefix=col, dtype=int)
return df
class CoordinateGetter(BaseEstimator, TransformerMixin):
def __init__(self):
pass
def fit(self, X, y=None):
return self
def transform(self, X):
df = X.copy()
if 'id' in df.columns:
df_giraffe = pd.read_csv('data/Giraffe.csv')
df_giraffe = df_giraffe[['propertyId', 'latitude', 'longitude']]
df_giraffe['id'] = df_giraffe['propertyId']
cols = df_giraffe.columns.tolist()
cols.remove('id')
new_order = ['id'] + cols
df_giraffe = df_giraffe[new_order]
df_giraffe = df_giraffe.drop(columns='propertyId')
df = df.merge(df_giraffe, on='id', how='left')
df = df.dropna(subset=['latitude', 'longitude'])
else :
print(f"DEBUG: PGEOCODE_CACHE_DIR from os.environ: '{os.environ.get('PGEOCODE_CACHE_DIR')}'")
print(f"DEBUG: XDG_CACHE_HOME from os.environ: '{os.environ.get('XDG_CACHE_HOME')}'")
print(f"DEBUG: os.path.expanduser('~'): '{os.path.expanduser('~')}'")
print(f"DEBUG: Current Working Directory: '{os.getcwd()}'")
# Try to create the /tmp/pgeocode directory to check permissions there
try:
temp_cache_dir = "/tmp/pgeocode"
os.makedirs(temp_cache_dir, exist_ok=True)
print(f"DEBUG: Successfully created/ensured existence of '{temp_cache_dir}'.")
except Exception as e:
print(f"DEBUG: Failed to create '{temp_cache_dir}': {e}")
nomi = pgeocode.Nominatim('be')
df['postCode'] = df['postCode'].astype(str)
unique_postcodes = df["postCode"].astype(str).unique()
geo_df = nomi.query_postal_code(list(unique_postcodes))
geo_df = geo_df[['postal_code', 'latitude', 'longitude']]
geo_df = geo_df.rename(columns={'postal_code': 'postCode'})
geo_df['postCode'] = geo_df['postCode'].astype(str)
df = df.merge(geo_df, on='postCode', how='left')
return df
class KDEKNNFeatureCreator(BaseEstimator, TransformerMixin):
def __init__(self, k=20):
self.k = k
self.scaler = StandardScaler()
self.knn = NearestNeighbors(n_neighbors=k)
self.train_prices = None
def fit(self, X, y=None):
if 'latitude' not in X.columns or 'longitude' not in X.columns:
print("Warning: Missing latitude/longitude columns")
return self
coords_scaled = self.scaler.fit_transform(X[['latitude', 'longitude']])
self.knn.fit(coords_scaled)
# Store training prices
self.train_prices = X['pricePerM2'].values
return self
def transform(self, X):
df = X.copy()
if 'latitude' not in df.columns or 'longitude' not in df.columns:
print("Warning: Missing latitude/longitude columns")
df['kde_price_per_m2_knn'] = np.nan
return df
coords_scaled = self.scaler.transform(df[['latitude', 'longitude']])
distances, indices = self.knn.kneighbors(coords_scaled)
kde_scores = []
invalid_kde_count = 0
for i in range(len(df)):
neighbor_idxs = indices[i]
# Use stored training prices for neighbors
neighbor_prices = self.train_prices[neighbor_idxs]
neighbor_prices = neighbor_prices[~np.isnan(neighbor_prices)]
if len(neighbor_prices) < 2:
kde_scores.append(np.nan)
invalid_kde_count += 1
continue
try:
kde = gaussian_kde(neighbor_prices)
value_to_evaluate = neighbor_prices.mean()
kde_score = kde(value_to_evaluate)[0]
if np.isfinite(kde_score):
kde_scores.append(kde_score)
else:
kde_scores.append(np.nan)
invalid_kde_count += 1
except Exception as e:
print(f"Error in KDE calculation for row {i}: {str(e)}")
kde_scores.append(np.nan)
invalid_kde_count += 1
df['kde_price_per_m2_knn'] = kde_scores
# Fill NaN values with median by region
df['kde_price_per_m2_knn'] = df['kde_price_per_m2_knn'].fillna(-1)
return df.drop(columns=['latitude', 'longitude'], errors='ignore')
class ColumnCleaner(BaseEstimator, TransformerMixin):
def __init__(self):
self.columns_to_drop = [
'id', 'postCode', 'buildingConstructionYear', 'type', 'locality', 'region',
'latitude', 'longitude', 'buildingConstructionYear'
]
def fit(self, X, y=None):
return self
def transform(self, X):
df = X.copy()
# Drop columns that are no longer needed
columns_to_drop = [col for col in self.columns_to_drop if col in df.columns]
df = df.drop(columns=columns_to_drop)
if 'pricePerM2' in df.columns:
df = df.drop(columns=['pricePerM2'])
# Ensure all remaining columns are numeric
non_numeric_cols = df.select_dtypes(include=['object', 'category']).columns
if len(non_numeric_cols) > 0:
# Convert any remaining categorical columns to numeric
for col in non_numeric_cols:
if col != 'price': # Don't encode the target variable
df[col] = pd.Categorical(df[col]).codes
# Reorganize columns to put price at the end
cols = df.columns.tolist()
if 'price' in cols:
cols.remove('price')
cols.append('price')
df = df[cols]
return df |