🧠 Girlfriend Prediction Using Logistic Regression

A learning project focused on understanding core machine learning concepts by implementing Logistic Regression both from scratch and using scikit-learn.

🎓 Purpose: This project is designed to learn and demonstrate understanding of:

• Gradient descent optimization from first principles
• Binary cross-entropy loss function
• Sklearn Pipelines for clean ML workflows

Uses synthetic data to focus on algorithm implementation rather than real-world prediction.

⚠️ Disclaimer: This is purely an educational project. Relationship outcomes depend on countless factors beyond what any model can capture!

---

📋 Table of Contents

• Overview
• Notebooks
• Dataset
• Technical Approach
• Results
• Installation
• Usage
• Project Structure
• Learning Outcomes

---

🎯 Overview

This project contains two implementations of Logistic Regression:

Notebook	Approach	Purpose
gf_pred_manual.ipynb	From scratch with NumPy	Understand gradient descent
gf_pred_sklearn.ipynb	Using sklearn Pipeline	Learn production-style workflow

Key Highlights

• ✨ Manual implementation with no ML libraries
• 📊 Custom gradient descent with 100,000 iterations
• 📈 Visualization of cost function convergence
• 🔧 Sklearn Pipeline for comparison

---

📊 Dataset

The dataset (indian_boys_gf_prediction_balanced.csv) contains 300 balanced samples with the following features:

Feature	Description	Range
age	Age in years	18-30
height_cm	Height in centimeters	155-190
income_lpa	Annual income (Lakhs/year)	1.5-20
fitness_level	Self-rated fitness score	1-10
confidence	Self-rated confidence score	1-10
social_media_hours	Daily social media usage (hours)	0.5-6.0
has_gf	Target variable	0 (No) / 1 (Yes)

Dataset Distribution

• Class 0 (No GF): ~50%
• Class 1 (Has GF): ~50%

---

🔬 Technical Approach

1. Feature Scaling (Standardization)

z = (x - μ) / σ

Ensures all features are on the same scale for faster gradient descent convergence.

2. Sigmoid Activation

σ(z) = 1 / (1 + e^(-z))

Maps linear output to probability between 0 and 1.

3. Binary Cross-Entropy Loss

J(W, B) = -(1/m) Σ [y·log(ŷ) + (1-y)·log(1-ŷ)]

4. Gradient Descent Update

W = W - α · (∂J/∂W)
B = B - α · (∂J/∂B)

Where α = 0.1 (learning rate)

---

📈 Results

Metric	Value
Training Accuracy	~65%
Test Accuracy	~54%
Iterations	100,000
Learning Rate	0.1

Observations

• Model converges successfully (cost function decreases)
• Moderate accuracy expected due to:
  • Inherently noisy/random nature of the target
  • Limited feature set
  • Small dataset size

---

🚀 Installation

Prerequisites

• Python 3.8+
• pip

Setup

# Clone the repository
git clone https://github.com/yourusername/logistic_regression_gf_pred_proj.git
cd logistic_regression_gf_pred_proj

# Install dependencies
pip install pandas numpy matplotlib scikit-learn jupyter

---

💻 Usage

Run the Notebooks

# Manual implementation (from scratch)
jupyter notebook gf_pred_manual.ipynb

# Sklearn implementation
jupyter notebook gf_pred_sklearn.ipynb

Make a Prediction

# Example: Predict for a new user
user_input = pd.DataFrame([{
    "age": 25,
    "height_cm": 180,
    "income_lpa": 12,
    "fitness_level": 10,
    "confidence": 7,
    "social_media_hours": 5
}])

# Scale using training statistics
scaled_input = (user_input.values - X_train.mean(axis=0)) / X_train.std(axis=0)

# Get probability
probability = sigmoid(np.dot(scaled_input, W) + B)
print(f"Probability of having GF: {probability[0][0]:.2%}")

---

📁 Project Structure

logistic_regression_gf_pred_proj/
├── gf_pred_manual.ipynb                    # From-scratch implementation
├── gf_pred_sklearn.ipynb                   # Sklearn Pipeline implementation
├── indian_boys_gf_prediction_balanced.csv  # Dataset
└── README.md                               # This file

---

🎓 Learning Outcomes

This project teaches:

• ✅ Logistic Regression theory and math
• ✅ Gradient Descent from first principles
• ✅ Binary Cross-Entropy loss function
• ✅ Feature scaling importance
• ✅ Train/Test split to prevent overfitting
• ✅ Sklearn Pipelines for clean workflows

---

🔮 Next Steps

For future projects with real-world data:

1. EDA: Exploratory Data Analysis before modeling
2. Feature Engineering: Create meaningful features
3. Cross-Validation: k-fold CV for robust evaluation
4. Hyperparameter Tuning: GridSearchCV / RandomizedSearchCV
5. Model Comparison: Try multiple algorithms

---

📚 Learning Outcomes

This project teaches:

• ✅ Logistic Regression theory and implementation
• ✅ Gradient Descent optimization
• ✅ Binary Cross-Entropy loss function
• ✅ Feature scaling importance
• ✅ Train/Test split to prevent overfitting
• ✅ Model evaluation metrics

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

🤝 Contributing

Contributions are welcome! Feel free to:

1. Fork the repository
2. Create a feature branch
3. Submit a pull request