🌟 Fitness Tracker Data Analysis & Prediction

Welcome to the Fitness Tracker Data Analysis & Prediction project! This project focuses on analyzing fitness tracking data and predicting Calories Burned using Deep Learning.

📚 Overview

We process a synthetic fitness dataset that includes various health and exercise-related parameters. We clean and preprocess the data, apply feature engineering, and train a Neural Network to predict the number of calories burned.

📊 Dataset Details

The dataset contains information on:

• Demographics: Age, Gender, Weight, Height
• Heart Rate Metrics: Max BPM, Average BPM, Resting BPM
• Fitness Stats: Workout Duration, Frequency, Experience Level
• Health Factors: Fat Percentage, BMI, Water Intake
• Workout Type
• Target Variable: Calories Burned

🎨 Data Preprocessing

To ensure high-quality data, we perform:

• Handling missing values (imputation & dropping critical NaNs)
• Feature engineering (BMI & Avg BPM calculations)
• Encoding categorical variables (One-Hot Encoding)
• Scaling numerical features using StandardScaler

💪 Model Architecture

We use a Deep Neural Network with the following structure:

• Dense(128, ReLU) + Dropout(0.3)
• Dense(128, ReLU) + Dropout(0.3)
• Dense(64, ReLU)
• Dense(1) (Output Layer)
• Optimizer: Adam
• Loss Function: Mean Squared Error (MSE)

📈 Training & Evaluation

The model is trained using early stopping for optimal performance. We evaluate using:

• Training & Validation Loss
• Learning Curves Visualization

🎨 Visualization

The learning curve is plotted to observe model performance over epochs:

plt.plot(history.history['val_loss'], label='Validation Loss')
plt.plot(history.history['loss'], label='Training Loss')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.show()

🎉 Enjoy exploring fitness insights and predictions!

Happy coding! 🚀