Sales Forecasting Report — Olist Dataset

This repository contains a comprehensive Jupyter Notebook, sales_forecasting_final.ipynb, which documents a complete data science workflow to build a sales forecasting model using the Olist e-commerce dataset.

Project Summary

The notebook demonstrates an end-to-end process for sales forecasting, including:

• Data ingestion and integration from multiple CSV tables.
• Data cleaning and validation (handling missing values, fixing inconsistencies, and removing duplicates).
• Exploratory Data Analysis (EDA) with descriptive statistics and visualizations to identify trends and outliers.
• Feature engineering for time-series and tabular modeling, including seasonality and aggregated metrics.
• Model training and evaluation using scikit-learn and XGBoost, with model selection and validation.

Key libraries used: pandas, numpy, matplotlib, seaborn, scikit-learn, xgboost.

Dataset

This project uses the Brazilian E-Commerce Public Dataset provided by Olist. The relevant CSV files are expected to be placed in data/full/ and include:

• olist_orders_dataset.csv — orders and timestamps (purchase, approval, delivery).
• olist_order_items_dataset.csv — order line items, product IDs, prices, shipping costs, seller IDs.
• olist_products_dataset.csv — product attributes (category, weight, dimensions).
• olist_customers_dataset.csv — customer identifiers and locations (ZIP code, city, state).
• olist_geolocation_dataset.csv — ZIP code latitude/longitude for geospatial analysis.
• olist_order_payments_dataset.csv — payment methods and installment details.
• olist_order_reviews_dataset.csv — customer ratings and review text.
• olist_sellers_dataset.csv — seller information and locations.

Notes:

• Raw data files are typically large and are excluded from this repository by .gitignore (see data/full/).
• Model artifacts (e.g., *.pkl) are also excluded from version control.

Usage

1. Create a Python environment and install dependencies:

   pip install -r requirements.txt

   If a requirements.txt file is not present, install the essentials:

   pip install pandas numpy matplotlib seaborn scikit-learn xgboost jupyter

2. Place the Olist CSV files in data/full/.

3. Launch Jupyter and open sales_forecasting_final.ipynb to run the analysis and reproduce the results.

License

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