Obesity Prediction using Machine Learning

This project predicts the obesity level of individuals based on their physical and lifestyle attributes using multiple machine learning models. It involves end-to-end data processing, exploratory data analysis (EDA), feature engineering, model evaluation, and hyperparameter tuning.

Open in Google Colab

📁 Dataset

• Source: The dataset is split into:
  • train_dataset.csv
  • test_dataset.csv
• Target Column: NObeyesdad — Multi-class label representing obesity levels.

🚀 Features Used

• Age
• Height
• Weight
• BMI (calculated)
• FCVC: Frequency of consuming vegetables
• NCP: Number of main meals
• CH2O: Daily water consumption
• FAF: Physical activity frequency
• TUE: Time using technology devices
• Gender, Family history, Food habits, Transportation methods, etc.

🧹 Data Preprocessing

• Calculated BMI from weight and height
• Handled missing values using mean/mode
• Removed duplicates
• One-Hot and Label Encoding for categorical data
• Outlier handling using IQR method
• Feature scaling using StandardScaler

📊 Exploratory Data Analysis (EDA)

• Boxplots, Distribution plots, Countplots
• Correlation heatmap
• Feature importance visualization using Random Forest

🧠 Models Implemented

• Logistic Regression
• Decision Tree
• Random Forest
• K-Nearest Neighbors (KNN)
• Neural Network (MLPClassifier)
• Support Vector Machine (SVM)
• Gradient Boosting Classifier
• Stacking Classifier (Ensemble)

🛠️ Model Evaluation Metrics

• Accuracy
• Precision (weighted)
• Recall (weighted)
• F1-Score (weighted)
• Confusion Matrix
• Classification Report

🔍 Hyperparameter Tuning

• GridSearchCV for:
  • Logistic Regression (C, penalty, solver)
  • KNN (n_neighbors, p, metric, algorithm)

🏆 Best Model (Stacking Ensemble)

Combines:

• Logistic Regression
• Tuned KNN
• Random Forest
• Decision Tree
• Neural Network
• SVM
• Gradient Boosting

Meta-model: Logistic Regression

💾 Model Saving

• StandardScaler saved as scaler.pkl
• Final SVM model saved as svm_model.pkl

📉 Visualizations

• Feature distributions
• Boxplots: Features vs Obesity Level
• Countplot of Obesity Classes
• Feature importance charts
• Correlation heatmap

📂 Project Structure

📦 Obesity-Prediction-ML ├── data/ │ ├── train_dataset.csv │ └── test_dataset.csv ├── models/ │ ├── svm_model.pkl │ └── scaler.pkl ├── obesity_prediction.ipynb / .py └── README.md