📱 Mobile Phones: What Are You Paying For?

This project investigates the relationship between smartphone features and their prices using various regression techniques. It aims to answer the question: Which features influence mobile phone pricing the most, and how accurately can we predict price using those features?

View the notebook for an in-depth review of techniques.

---

🔧 Setup

Install the required packages:

pip install -r requirements.txt

Or install individually:

pip install numpy pandas matplotlib seaborn scikit-learn statsmodels kaggle jupyter ipykernel

---

🎯 Objective

Explore and compare multiple regression models to determine the most effective one for predicting mobile phone prices.

---

📊 Workflow

1. 📁 Dataset Overview

• Load and describe the dataset
• Understand structure, datatypes, and missing values

2. 📈 Exploratory Data Analysis (EDA)

• Visualize the distribution of the target variable (Price)
• Analyze individual features
• Explore feature-target relationships
• Check for multicollinearity using VIF and correlation heatmaps

3. 🧹 Data Preprocessing

• Handle missing or inconsistent values
• Normalize / transform variables as needed
• Split into training and test sets

4. 🤖 Model Training & Evaluation

Models Explored:

• 🔹 Ordinary Least Squares (OLS) Regression
• 🔹 Random Forest Regressor
• 🔹 Ridge Regression (for multicollinearity)

Evaluation Metrics:

• 📉 Mean Absolute Error (MAE)
• 📉 Mean Squared Error (MSE)
• 📈 R-squared (R²)
• 📊 Variance Inflation Factor (VIF) for multicollinearity detection
• 📉 Actual vs Predicted scatter plots

---

🧠 Key Insights

• Dataset: 807 mobile phones with 8 features (Ratings, RAM, ROM, Mobile_Size, Primary_Cam, Selfi_Cam, Battery_Power, Price)
• Data Preprocessing: Applied log transformation to stabilize price variance and handle right-skewed distribution
• Multicollinearity: High VIF values detected in Ratings (40.5), Battery_Power (18.4), and Primary_Cam (18.0)
• Model Performance:
  • OLS Regression: R² = 0.67
  • Random Forest: R² = 0.95 (best performance)
• Key Predictors: Ratings (0.69), ROM (0.61), and Battery_Power (0.55) show strongest correlation with price
• Surprising Findings: Primary_Cam shows negative correlation (-0.27) with price

---

📌 Future Work

• Feature selection and dimensionality reduction (PCA)
• Deploy model via a Flask app or Streamlit dashboard
• Integration with real-time pricing APIs for prediction

---

🛠️ Technologies & Tools

Core Libraries

Development Environment

Key Technologies

• NumPy - Numerical computations and array operations
• Pandas - Data manipulation and analysis
• Matplotlib - Data visualization and plotting
• Seaborn - Statistical data visualization
• Scikit-learn - Machine learning algorithms and metrics
• Statsmodels - Statistical modeling and analysis

Statistical Analysis

• Variance Inflation Factor (VIF) - Multicollinearity detection
• Correlation Analysis - Feature relationship analysis
• Log Transformation - Data normalization
• Train-Test Split - Model validation

Machine Learning Models

• Ordinary Least Squares (OLS) - Linear regression with statistical inference
• Random Forest Regressor - Ensemble learning method
• Ridge Regression - Regularized linear regression
• StandardScaler - Feature scaling and normalization

---

🙌 Acknowledgments

• Dataset: Mobile Phone Price Prediction by Ganjerlawrence on Kaggle
• Visualization powered by Matplotlib and Seaborn
• Models built with Scikit-learn and Statsmodels

---

📂 License

This project is open-source under the MIT License.