sleep-lifecycle-predictor / train_model.py
varshagoswami01's picture
Upload 5 files
7075fa2 verified
Raw History Blame Contribute Delete
5.46 kB
"""
train_model.py
==============
Trains a Random Forest Classifier to predict healthy vs unhealthy lifestyle status
using the Sleep Health and Lifestyle dataset.
Saves the trained pipeline model as lifestyle_model.pkl.
"""
import os
import joblib
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import OrdinalEncoder
from sklearn.pipeline import Pipeline
from sklearn.metrics import accuracy_score, classification_report, confusion_matrix
DATASET_PATH = "Sleep_health_and_lifestyle_dataset.csv"
MODEL_PATH = "lifestyle_model.pkl"
def load_and_preprocess_data(dataset_path: str):
"""Loads the dataset, engineer features and target variable."""
if not os.path.exists(dataset_path):
raise FileNotFoundError(f"Dataset not found at {dataset_path}")
print(f"Loading dataset from: {dataset_path} ...")
df = pd.read_csv(dataset_path)
print(f"Dataset shape: {df.shape}")
# 1. Define Binary Target Variable 'Is_Healthy'
# Healthy (1) if:
# - Sleep Duration >= 7.0 hours (recommended minimum for adults)
# - Quality of Sleep >= 7 (on 1-10 scale)
# - Physical Activity Level >= 60 (minutes/day)
# - Stress Level <= 5 (low to moderate stress)
# Otherwise Unhealthy (0)
df["Is_Healthy"] = (
(df["Sleep Duration"] >= 7.0)
& (df["Quality of Sleep"] >= 7)
& (df["Physical Activity Level"] >= 60)
& (df["Stress Level"] <= 5)
).astype(int)
target_counts = df["Is_Healthy"].value_counts().to_dict()
print(f"Target distribution: 0 (Unhealthy): {target_counts.get(0, 0)}, 1 (Healthy): {target_counts.get(1, 0)}")
# 2. Preprocess Blood Pressure into numeric Systolic_BP and Diastolic_BP
df[["Systolic_BP", "Diastolic_BP"]] = (
df["Blood Pressure"].str.split("/", expand=True).astype(int)
)
# 3. Exclude BMI Category (to keep features clean and unconfounded)
# Also drop Person ID (identifier) and Sleep Disorder (outcome diagnosis with NaNs)
feature_cols = [
"Gender",
"Age",
"Occupation",
"Sleep Duration",
"Quality of Sleep",
"Physical Activity Level",
"Stress Level",
"Heart Rate",
"Daily Steps",
"Systolic_BP",
"Diastolic_BP",
]
X = df[feature_cols].copy()
y = df["Is_Healthy"].copy()
return X, y, feature_cols
def build_pipeline(categorical_cols, numerical_cols):
"""Creates a Scikit-Learn Pipeline with preprocessor and RandomForestClassifier."""
preprocessor = ColumnTransformer(
transformers=[
(
"cat",
OrdinalEncoder(
handle_unknown="use_encoded_value", unknown_value=-1
),
categorical_cols,
),
("num", "passthrough", numerical_cols),
]
)
pipeline = Pipeline(
steps=[
("preprocessor", preprocessor),
(
"classifier",
RandomForestClassifier(
n_estimators=100,
max_depth=6,
random_state=42,
class_weight="balanced",
),
),
]
)
return pipeline
def main():
print("=" * 60)
print(" Sleep Health & Lifestyle AI Classifier - Training Pipeline ")
print("=" * 60)
X, y, feature_cols = load_and_preprocess_data(DATASET_PATH)
categorical_cols = ["Gender", "Occupation"]
numerical_cols = [col for col in feature_cols if col not in categorical_cols]
print(f"Categorical features: {categorical_cols}")
print(f"Numerical features: {numerical_cols}")
# Train / Test Split for Evaluation
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=0.2, random_state=42, stratify=y
)
print(f"\nTraining set size: {len(X_train)} samples")
print(f"Testing set size: {len(X_test)} samples")
# Evaluate Model
eval_pipeline = build_pipeline(categorical_cols, numerical_cols)
eval_pipeline.fit(X_train, y_train)
y_pred = eval_pipeline.predict(X_test)
accuracy = accuracy_score(y_test, y_pred)
print("\n" + "=" * 60)
print(" EVALUATION RESULTS (Holdout Test Set) ")
print("=" * 60)
print(f"Accuracy Score: {accuracy * 100:.2f}%\n")
print("Classification Report:")
print(
classification_report(
y_test, y_pred, target_names=["Unhealthy (0)", "Healthy (1)"]
)
)
print("Confusion Matrix:")
print(confusion_matrix(y_test, y_pred))
# Train final model on full dataset for production deployment
print("\n" + "=" * 60)
print(" Training Final Model on Full Dataset for Deployment ")
print("=" * 60)
final_pipeline = build_pipeline(categorical_cols, numerical_cols)
final_pipeline.fit(X, y)
# Save trained pipeline model
joblib.dump(final_pipeline, MODEL_PATH)
print(f"Successfully saved trained model to: {MODEL_PATH}")
# Verify model artifact can be reloaded
loaded_model = joblib.load(MODEL_PATH)
sample_pred = loaded_model.predict(X.iloc[:2])
print(f"Verification test prediction on first 2 samples: {sample_pred}")
print("Model training and packaging completed successfully!")
if __name__ == "__main__":
main()