import pandas as pd
import numpy as np
from sklearn.compose import ColumnTransformer
from sklearn.pipeline import Pipeline
from sklearn.impute import SimpleImputer
from sklearn.preprocessing import StandardScaler, OneHotEncoder, LabelEncoder
from sklearn.linear_model import LogisticRegression
from sklearn.inspection import permutation_importance
from sklearn.model_selection import train_test_split
from sklearn.metrics import f1_score
import joblib
# Load the dataset
data = pd.read_csv('data.csv')
# Defining column types
id_col = 'customerID'
num_cols = ['SeniorCitizen', 'tenure', 'MonthlyCharges', 'TotalCharges']
cat_cols = [
'gender', 'Partner', 'Dependents',
'PhoneService', 'MultipleLines', 'InternetService',
'OnlineSecurity', 'OnlineBackup', 'DeviceProtection',
'TechSupport', 'StreamingTV', 'StreamingMovies',
'Contract', 'PaperlessBilling', 'PaymentMethod'
]
# Define X (features) and y (target variable)
X = data.drop(['Churn', id_col], axis=1)
y = data['Churn']
# Convert labels to numeric form
label_encoder = LabelEncoder()
y = label_encoder.fit_transform(y)
# Split the dataset into training and test sets
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=0.2, random_state=42
)
# Define transformers
numeric_transformer = Pipeline(steps=[
('imputer', SimpleImputer(strategy='mean')),
('scaler', StandardScaler())
])
categorical_transformer = Pipeline(steps=[
('imputer', SimpleImputer(strategy='most_frequent')),
('encoder', OneHotEncoder(handle_unknown='error', drop='first'))
])
# Create the column transformer
preprocessor = ColumnTransformer(transformers=[
('numeric', numeric_transformer, num_cols),
('categorical', categorical_transformer, cat_cols)
])
# Create the pipeline with the preprocessor and classifier
pipeline = Pipeline(steps=[
('preprocessor', preprocessor),
('classifier', LogisticRegression())
])
# Fit the pipeline
pipeline.fit(X_train, y_train)
# Calculate feature importance using permutation
result = permutation_importance(
pipeline, X_test, y_test, scoring='f1', n_repeats=10, random_state=42
)
importance_scores = result.importances_mean
for col, score in zip([col for col in list(X)], importance_scores):
print(f"{col}: {score}")
# Export the pipeline using joblib
joblib.dump(pipeline, 'pipeline.joblib')
# Get a new observation for testing
new_data = pd.read_csv('new_data.csv')
new_data.drop(id_col, axis=1, inplace=True)
# Apply the pipeline to the new observation
prediction = pipeline.predict(new_data)
final_prediction=label_encoder.inverse_transform(prediction)
print("Final prediction:", final_prediction)