train.py

import joblib
import pandas as pd
import numpy as np
from sklearn.preprocessing import StandardScaler, OneHotEncoder
from sklearn.compose import make_column_transformer
from sklearn.impute import SimpleImputer
from sklearn.pipeline import Pipeline
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestRegressor
from sklearn.metrics import mean_squared_error, r2_score, mean_absolute_error

# Load the dataset
df = pd.read_csv("insurance.csv")

# Define features and target
target = 'charges'
numerical_features = ['age', 'bmi', 'children']
categorical_features = ['sex', 'smoker', 'region']

print("Creating data subsets")

X = df[numerical_features + categorical_features]
y = df[target]

Xtrain, Xtest, ytrain, ytest = train_test_split(
    X, y,
    test_size=0.2,
    random_state=42
)

# Define the numerical and categorical pipelines
numerical_pipeline = Pipeline([
    ('imputer', SimpleImputer(strategy='median')),
    ('scaler', StandardScaler())
])

categorical_pipeline = Pipeline([
    ('imputer', SimpleImputer(strategy='most_frequent')),
    ('onehot', OneHotEncoder(handle_unknown='ignore'))
])

preprocessor = make_column_transformer(
    (numerical_pipeline, numerical_features),
    (categorical_pipeline, categorical_features)
)

# Define the Random Forest model with the best parameters
model_random_forest = RandomForestRegressor(
    n_estimators=125,
    min_samples_split=3,
    min_samples_leaf=4,
    max_depth=25,
    random_state=42,
    n_jobs=-1
)

print("Estimating Best Model Pipeline")

model_pipeline = Pipeline([
    ('preprocessor', preprocessor),
    ('regressor', model_random_forest)
])

# Train the model
model_pipeline.fit(Xtrain, ytrain)

# Predict on the test set
y_pred = model_pipeline.predict(Xtest)

# Calculate evaluation metrics
mae = mean_absolute_error(ytest, y_pred)
mse = mean_squared_error(ytest, y_pred)
rmse = np.sqrt(mse)
r2 = r2_score(ytest, y_pred)

print("Logging Metrics")
print(f"Mean Absolute Error (MAE): {mae}")
print(f"Mean Squared Error (MSE): {mse}")
print(f"Root Mean Squared Error (RMSE): {rmse}")
print(f"R-squared (R²): {r2}")

print("Serializing Model")

# Save the model to a file
saved_model_path = "random_forest_pipeline.pkl"
joblib.dump(model_pipeline, saved_model_path)
print(f"Model saved as {saved_model_path}")