File size: 17,398 Bytes
aba2f7b | 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 | import sys
import os
import numpy as np
import pandas as pd
from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import OrdinalEncoder, StandardScaler
from sklearn.model_selection import train_test_split
from src.exception.exception import CustomException # Assuming you have this
from src.logging.logger import logging # Assuming you have this
from src.utils.main_utils.utils import save_numpy_array_data,save_object
from src.entity.artifact_entity import (
DataTransformationArtifact,
DataValidationArtifact
)
from src.entity.config_entity import DataTransformationConfig
# @dataclass
# class DataTransformationConfig:
# transformed_object_file_path: str = os.path.join("artifacts", "preprocessor.pkl")
# transformed_train_file_path: str = os.path.join("artifacts", "transformed_train.npy")
# transformed_test_file_path: str = os.path.join("artifacts", "transformed_test.npy")
# label_encoders_file_path: str = os.path.join("artifacts", "label_encoders.pkl")
# @dataclass
# class DataTransformationArtifact:
# transformed_object_file_path: str
# transformed_train_file_path: str
# transformed_test_file_path: str
# label_encoders_file_path: str
class DataTransformation:
def __init__(self,data_validation_artifact:DataValidationArtifact,
data_transformation_config:DataTransformationConfig):
try:
self.data_validation_artifact:DataValidationArtifact=data_validation_artifact
self.data_transformation_config:DataTransformationConfig=data_transformation_config
except Exception as e:
raise CustomException(e,sys)
@staticmethod
def read_data(file_path) -> pd.DataFrame:
try:
return pd.read_csv(file_path)
except Exception as e:
raise CustomException(e, sys)
def get_data_transformer_object(self, categorical_features, numeric_cols, ordinal_categories):
try:
logging.info("Creating data transformer object...")
# Define the ordinal encoder with the given category order
ordinal_encoder = OrdinalEncoder(
categories=[ordinal_categories[col] for col in categorical_features],
dtype=int,
handle_unknown='use_encoded_value',
unknown_value=-1, # Fix: Use a float value instead of np.nan
# dtype=np.float64 # Fix: Ensure dtype is float to match NaNs
)
# Standard Scaler for numerical features
scaler = StandardScaler()
# Define transformations
cat_transformers = [("ordinal", ordinal_encoder, categorical_features)]
num_transformers = [("scaler", scaler, numeric_cols)]
# Combine transformers using ColumnTransformer
preprocessor = ColumnTransformer(
transformers=cat_transformers + num_transformers,
remainder='passthrough'
)
logging.info("Data transformer object created successfully.")
return preprocessor
except Exception as e:
raise CustomException(e, sys)
def perform_feature_engineering(self, df):
# All of your complex feature engineering logic goes here
df['InvoiceDate'] = pd.to_datetime(df['InvoiceDate'], errors='coerce')
df= df[(df['Quantity'] >= 0) & (df['UnitPrice'] >= 0)]
df = df.copy() # Ensure df is not a slice
df.loc[:, 'hour'] = df['InvoiceDate'].dt.hour
df.loc[:, 'weekday'] = df['InvoiceDate'].dt.weekday
df.loc[:, 'month'] = df['InvoiceDate'].dt.month
df.loc[:, 'week'] = df['InvoiceDate'].dt.isocalendar().week
df.loc[:, 'total_sales'] = df['UnitPrice'] * df['Quantity']
def calculate_peak_periods(data, group_by_column, top_n_percent=20):
period_sales = data.groupby(group_by_column)['total_sales'].sum().reset_index()
period_sales_sorted = period_sales.sort_values(by='total_sales', ascending=False)
num_peak_periods = int((top_n_percent / 100) * len(period_sales_sorted))
peak_periods = period_sales_sorted.head(num_peak_periods)[group_by_column].tolist()
return peak_periods
peak_months = calculate_peak_periods(df, 'month')
peak_weeks = calculate_peak_periods(df, 'week')
peak_weekdays = calculate_peak_periods(df, 'weekday')
peak_hours = calculate_peak_periods(df, 'hour')
print("Peak Months:", peak_months)
print("Peak Weeks:", peak_weeks)
print("Peak Weekdays:", peak_weekdays)
print("Peak Hours:", peak_hours)
def assign_peak_level_combined(row):
peak_count = sum([
row['month'] in peak_months,
row['week'] in peak_weeks,
row['weekday'] in peak_weekdays,
row['hour'] in peak_hours
])
return ['Very Low', 'Low', 'Medium', 'High', 'Very High'][peak_count]
df.loc[:, 'peak_period_level'] = df.apply(assign_peak_level_combined, axis=1)
product_demand = df.groupby('StockCode')['Quantity'].sum().reset_index()
product_demand.rename(columns={'Quantity': 'total_quantity'}, inplace=True)
high_demand_threshold = np.percentile(product_demand['total_quantity'], 80)
low_demand_threshold = np.percentile(product_demand['total_quantity'], 20)
def assign_demand_level(total_quantity):
if total_quantity > high_demand_threshold:
return 'High'
elif total_quantity < low_demand_threshold:
return 'Low'
else:
return 'Medium'
product_demand['overall_demand_level'] = product_demand['total_quantity'].apply(assign_demand_level)
df = df.merge(product_demand[['StockCode', 'overall_demand_level']], on='StockCode')
reference_date = df['InvoiceDate'].max()
print(reference_date)
rfm = df.groupby('CustomerID').agg({
'InvoiceDate': lambda x: (reference_date - x.max()).days,
'InvoiceNo': 'count',
'UnitPrice': 'sum'
}).reset_index()
rfm.columns = ['CustomerID', 'Recency', 'Frequency', 'Monetary']
# rfm['RecencySegment'] = pd.qcut(rfm['Recency'], 4, labels=['High', 'Medium', 'Low', 'Very Low'], duplicates='drop')
# rfm['FrequencySegment'] = pd.qcut(rfm['Frequency'], 4, labels=['Very Low', 'Low', 'Medium', 'High'], duplicates='drop')
# rfm['MonetarySegment'] = pd.qcut(rfm['Monetary'], 4, labels=['Very Low', 'Low', 'Medium', 'High'], duplicates='drop')
num_bins = 4
recency_bins = pd.qcut(rfm['Recency'], num_bins, duplicates='drop')
frequency_bins = pd.qcut(rfm['Frequency'], num_bins, duplicates='drop')
monetary_bins = pd.qcut(rfm['Monetary'], num_bins, duplicates='drop')
rfm['RecencySegment'] = pd.qcut(rfm['Recency'], num_bins, labels=[f"Q{i+1}" for i in range(len(recency_bins.cat.categories))], duplicates='drop')
rfm['FrequencySegment'] = pd.qcut(rfm['Frequency'], num_bins, labels=[f"Q{i+1}" for i in range(len(frequency_bins.cat.categories))], duplicates='drop')
rfm['MonetarySegment'] = pd.qcut(rfm['Monetary'], num_bins, labels=[f"Q{i+1}" for i in range(len(monetary_bins.cat.categories))], duplicates='drop')
df = df.merge(rfm[['CustomerID', 'RecencySegment','FrequencySegment','MonetarySegment']], on='CustomerID', how='left')
all_countries_stats = df.groupby('Country').agg(
total_sales=('total_sales', 'sum'),
avg_unit_price=('UnitPrice', 'mean')
).reset_index()
countries_excluding_uk_stats = all_countries_stats[all_countries_stats['Country'] != 'United Kingdom']
high_demand_threshold_excluding_uk = np.percentile(countries_excluding_uk_stats['total_sales'], 80)
low_demand_threshold_excluding_uk = np.percentile(countries_excluding_uk_stats['total_sales'], 20)
high_demand_threshold_including_uk = np.percentile(all_countries_stats['total_sales'], 80)
low_demand_threshold_including_uk = np.percentile(all_countries_stats['total_sales'], 20)
def demand_level(country, total_sales):
high_threshold, low_threshold = (
(high_demand_threshold_including_uk, low_demand_threshold_including_uk)
if country == 'United Kingdom'
else (high_demand_threshold_excluding_uk, low_demand_threshold_excluding_uk)
)
return 'High' if total_sales > high_threshold else 'Low' if total_sales < low_threshold else 'Medium'
all_countries_stats['country_purchasing_power'] = all_countries_stats.apply(
lambda x: demand_level(x['Country'], x['total_sales']), axis=1
)
df = df.merge(all_countries_stats[['Country', 'country_purchasing_power']], on='Country', how='left')
grouped_sales = df.groupby(['StockCode', 'Country'])['total_sales'].sum().reset_index()
def calculate_thresholds(group):
thresholds = {
'very_low_threshold': group['total_sales'].quantile(0.10),
'low_threshold': group['total_sales'].quantile(0.25),
'medium_threshold': group['total_sales'].quantile(0.50),
'high_threshold': group['total_sales'].quantile(0.75),
'very_high_threshold': group['total_sales'].quantile(0.90),
}
return pd.Series(thresholds)
thresholds = grouped_sales.groupby('StockCode').apply(calculate_thresholds).reset_index()
grouped_sales = grouped_sales.merge(thresholds, on='StockCode')
def assign_level(row):
if row['total_sales'] <= row['very_low_threshold']:
return 'Very Low'
elif row['total_sales'] <= row['low_threshold']:
return 'Low'
elif row['total_sales'] <= row['medium_threshold']:
return 'Medium'
elif row['total_sales'] <= row['high_threshold']:
return 'High'
else:
return 'Very High'
grouped_sales['sales_level_by_country'] = grouped_sales.apply(assign_level, axis=1)
df = df.merge(grouped_sales[['StockCode', 'Country', 'sales_level_by_country']],on=['StockCode','Country'],how='left')
def adjust_unit_price(row):
price_adjustment = 0
if row['peak_period_level'] == 'High':
price_adjustment += 0.10
elif row['peak_period_level'] == 'Very High':
price_adjustment += 0.20
elif row['peak_period_level'] == 'Low':
price_adjustment -= 0.05
elif row['peak_period_level'] == 'Very Low':
price_adjustment -= 0.10
if row['overall_demand_level'] == 'High':
price_adjustment += 0.15
elif row['overall_demand_level'] == 'Medium':
price_adjustment += 0.05
elif row['overall_demand_level'] == 'Low':
price_adjustment -= 0.10
if row['RecencySegment'] == 'High':
price_adjustment += 0.05
elif row['RecencySegment'] == 'Medium':
price_adjustment += 0.03
elif row['RecencySegment'] == 'Low' or row['RecencySegment'] == 'Very Low':
price_adjustment -= 0.05
if row['FrequencySegment'] == 'High':
price_adjustment += 0.10
elif row['FrequencySegment'] == 'Medium':
price_adjustment += 0.05
elif row['FrequencySegment'] == 'Low' or row['FrequencySegment'] == 'Very Low':
price_adjustment -= 0.05
if row['MonetarySegment'] == 'High':
price_adjustment += 0.10
elif row['MonetarySegment'] == 'Medium':
price_adjustment += 0.05
elif row['MonetarySegment'] == 'Low' or row['MonetarySegment'] == 'Very Low':
price_adjustment -= 0.05
if row['country_purchasing_power'] == 'High':
price_adjustment += 0.08
elif row['country_purchasing_power'] == 'Medium':
price_adjustment += 0.04
elif row['country_purchasing_power'] == 'Low':
price_adjustment -= 0.08
if row['sales_level_by_country'] == 'Very High':
price_adjustment += 0.15
elif row['sales_level_by_country'] == 'High':
price_adjustment += 0.12
elif row['sales_level_by_country'] == 'Medium':
price_adjustment += 0.05
elif row['sales_level_by_country'] == 'Low':
price_adjustment -= 0.10
elif row['sales_level_by_country'] == 'Very Low':
price_adjustment -= 0.15
new_unit_price = row['UnitPrice'] * (1 + price_adjustment)
return new_unit_price
df['adjusted_unit_price'] = df.apply(adjust_unit_price, axis=1)
return df
def split_train_test(self, X: pd.DataFrame, y: pd.Series, test_size: float = 0.2, random_state: int = 42):
logging.info("Splitting dataset into train and test sets...")
train_X, test_X, train_y, test_y = train_test_split(X, y, test_size=test_size, random_state=random_state)
train_df = pd.concat([train_X, train_y], axis=1)
test_df = pd.concat([test_X, test_y], axis=1)
logging.info(f"Train shape: {train_df.shape}, Test shape: {test_df.shape}")
return train_df, test_df
def initiate_data_transformation(self) -> DataTransformationArtifact:
logging.info("Entered initiate_data_transformation method of DataTransformation class")
try:
logging.info("Starting data transformation")
df = self.read_data(self.data_validation_artifact.valid_data_file_path)
df = self.perform_feature_engineering(df)
categorical_features = ['peak_period_level', 'overall_demand_level', 'RecencySegment',
'FrequencySegment', 'MonetarySegment', 'country_purchasing_power', 'sales_level_by_country']
numeric_cols = ['Quantity', 'UnitPrice', 'total_sales']
target_column = 'adjusted_unit_price'
X = df.drop(columns=[target_column]) # Features (all columns except target)
y = df[target_column]
train_df, test_df = self.split_train_test(X, y)
print("Train columns:", train_df.columns)
print("Test columns:", test_df.columns)
print("Expected categorical features:", categorical_features)
print("Expected numeric features:", numeric_cols)
ordinal_categories = {
"peak_period_level": ["Very Low", "Low", "Medium", "High", "Very High"],
"overall_demand_level": ["Low", "Medium", "High"],
"RecencySegment": ["Very Low", "Low", "Medium", "High"],
"FrequencySegment": ["Very Low", "Low", "Medium", "High"],
"MonetarySegment": ["Very Low", "Low", "Medium", "High"],
"country_purchasing_power": ["Low", "Medium", "High"],
"sales_level_by_country": ["Very Low", "Low", "Medium", "High", "Very High"]
}
transformer = self.get_data_transformer_object(
categorical_features=list(ordinal_categories.keys()),
numeric_cols=['Quantity', 'UnitPrice', 'total_sales'],
ordinal_categories=ordinal_categories
)
input_feature_train_df = train_df[categorical_features + ['UnitPrice', 'Quantity','total_sales']]
target_feature_train_df = train_df[target_column]
input_feature_test_df = test_df[categorical_features + ['UnitPrice', 'Quantity','total_sales']]
target_feature_test_df = test_df[target_column]
logging.info("Applying preprocessing object to datasets...")
input_feature_train_arr = transformer.fit_transform(input_feature_train_df)
input_feature_test_arr = transformer.transform(input_feature_test_df)
train_arr = np.c_[input_feature_train_arr, np.array(target_feature_train_df)]
test_arr = np.c_[input_feature_test_arr, np.array(target_feature_test_df)]
save_numpy_array_data(self.data_transformation_config.transformed_train_file_path, array=train_arr)
save_numpy_array_data(self.data_transformation_config.transformed_test_file_path, array=test_arr)
save_object(self.data_transformation_config.transformed_object_file_path, transformer)
save_object( "final_model/preprocessor.pkl", transformer)
data_transformation_artifact = DataTransformationArtifact(
transformed_object_file_path=self.data_transformation_config.transformed_object_file_path,
transformed_train_file_path=self.data_transformation_config.transformed_train_file_path,
transformed_test_file_path=self.data_transformation_config.transformed_test_file_path,
)
return data_transformation_artifact
except Exception as e:
raise CustomException(e, sys) |