CPI Prediction with Ridge Regression

This project implements a Ridge Regression model to predict the Consumer Price Index (CPI) based on economic indicators. The model has been fine-tuned using GridSearchCV and evaluated on real economic data.

---

Installation

1. Clone the Repository

git clone <repository-url>
cd cpi-prediction

2. Install Dependencies

pip install -r requirements.txt

---

Usage

1. Data Preprocessing

Prepare and clean the dataset for model training:

python scripts/preprocess_data.py

• Loads raw CPI and macroeconomic data.
• Handles missing values and feature engineering.
• Saves the cleaned dataset to ./data/processed/feature_engineered_macro_data.csv.

2. Model Training

Train the Ridge Regression model using fine-tuned hyperparameters:

python scripts/train_model.py

• Uses GridSearchCV to find the optimal Ridge regression parameters.
• Saves the trained model to ./models/ridge_model.pkl.
• Computes Mean Absolute Error (MAE) and R² Score.
• Generates plots comparing Actual vs. Predicted CPI.

---

Files and Scripts

1. preprocess_data.py

• Cleans and prepares macroeconomic data for modeling.
• Saves processed data to the ./data/processed/ folder.

2. train_model.py

• Trains a Ridge Regression model.
• Uses GridSearchCV for hyperparameter tuning.
• Saves the trained model to ./models/ridge_model.pkl.
• Evaluates the model’s performance on a test dataset.
• Outputs:
  • MAE (Mean Absolute Error)
  • R² Score
  • Actual vs. Predicted CPI Scatter Plot

---

Results

Final Model Performance

• Fine-Tuned Ridge Alpha: 1.5
• Mean Absolute Error (MAE): 0.5444
• R² Score: 0.9998

Generated Plots

• Predicted vs. Actual CPI Scatter Plot
• Residual Analysis for Model Performance

---

Dependencies

• Python 3.12
• scikit-learn (Machine Learning)
• pandas (Data Handling)
• numpy (Numerical Computation)
• matplotlib (Visualization)
• seaborn (Enhanced Visualization)
• joblib (Model Saving and Loading)

To see the full list of dependencies, refer to requirements.txt.

---

Next Steps

• Implement Lasso Regression for feature selection.
• Experiment with Deep Learning models for CPI prediction.
• Automate data updates using live economic feeds.