Sales Forecasting Project

Welcome to the Sales Forecasting Project, an end-to-end implementation of a machine learning pipeline designed to predict daily sales for retail stores. This project showcases data preprocessing, feature engineering, model training, hyperparameter tuning, and model evaluation with advanced visualizations.

---

Table of Contents

1. Project Overview
2. Dataset Description
3. Project Workflow
4. Requirements
5. Setup and Installation
6. Usage
7. Results
8. File Structure
9. Contributing
10. License

---

Project Overview

This project aims to develop a Sales Forecasting Model for retail stores using XGBoost regression. The pipeline covers:

• Data Preprocessing: Cleaning and transforming raw data for better feature representation.
• Feature Engineering: Enhancing the dataset with meaningful features.
• Model Training and Tuning: Hyperparameter tuning using grid search and XGBoost's built-in cross-validation.
• Evaluation: Model evaluation using metrics such as RMSE and MAE.
• Visual Insights: SHAP visualizations and feature importance to interpret the model's predictions.

---

Dataset Description

The dataset contains daily sales data for retail stores, including features such as:

• Store Information: Store type, promotions, etc.
• Date: Daily sales and customer information.
• Holidays: School holidays and state holidays.

The raw dataset is located in the data/raw folder.

---

Project Workflow

1. Exploratory Data Analysis (EDA):

   • Conducted in the notebooks/eda.ipynb notebook.
   • Gained insights into trends, seasonality, and feature distributions.

2. Data Preprocessing:

   • Cleaned and transformed raw data.
   • Encoded categorical variables.

3. Feature Engineering:

   • Enhanced data with engineered features like year, month, week, and holidays.

4. Model Training and Tuning:

   • Trained a baseline model for comparison.
   • Used XGBoost for hyperparameter tuning.

5. Model Evaluation:

   • Evaluated the final model using RMSE and MAE.
   • Visualized SHAP summary plots and feature importance.

---

Requirements

• Python 3.10 or higher
• Required libraries are listed in requirements.txt. Install them using:

  pip install -r requirements.txt
  
  

Setup and Installation

1. Clone the Repository:

   git clone https://github.com/DataStatsMohith/Sales-Forecasting.git
   cd Sales-Forecasting

2. Set Up Virtual Environment:

   python -m venv sales_forecasting_env
   source sales_forecasting_env/bin/activate  # Mac/Linux
   sales_forecasting_env\Scripts\activate    # Windows

3. Install Dependencies:

   pip install -r requirements.txt

4. Run the Project:

   • Preprocess data:

     python scripts/preprocess_data.py

   • Engineer features:

     python scripts/feature_engineering.py

   • Train and tune the model:

     python scripts/xgboost_tuning.py

---

Usage

• Run Scripts: Use the provided Python scripts in the scripts folder to preprocess data, train models, and generate predictions.

• EDA Notebook: Explore insights in notebooks/eda.ipynb.

---

Results

• Final Model:

  • RMSE: 400.96
  • MAE: 245.70

• Feature Importance:

• SHAP Analysis:

---

File Structure

Sales-Forecasting/
│
├── data/
│   ├── raw/                    # Raw dataset
│   ├── processed/              # Processed and engineered data
│   ├── X_train.csv             # Training features
│   ├── X_val.csv               # Validation features
│   ├── y_train.csv             # Training target
│   ├── y_val.csv               # Validation target
│
├── notebooks/
│   ├── eda.ipynb               # Exploratory Data Analysis notebook
│
├── plots/                      # Visualizations
│   ├── feature_importance.png
│   ├── shap_summary_plot.png
│   ├── shap_force_plot.html
│
├── scripts/                    # Python scripts for the project
│   ├── preprocess_data.py
│   ├── feature_engineering.py
│   ├── model_preparation.py
│   ├── xgboost_tuning.py
│   ├── utils.py
│
├── models/                     # Trained models
│   ├── final_xgb_model.pkl
│
├── requirements.txt            # Project dependencies
├── README.md                   # Project documentation

---

Contributing

Contributions are welcome! Feel free to fork this repository and submit a pull request with your changes.

---

License

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