How rusirudananga/ml_cw works
This repository is a supermarket sales and quantity forecasting application, with a Streamlit interface as its user-facing runtime. Users select a prediction page, provide a date, department and store, and request a quantity or net-sales forecast; the app loads a saved XGBoost model and displays predictions. The documented offline workflow covers data preparation, EDA, model development and saving. Notebook internals and some app wiring were not sampled, so those relationships are shown only where documented.
Main entry points: run_pipeline.ipynb Pipeline orchestration · index.py Streamlit app · transformation_util.ipynb Feature preparation · qty_pred_wdepartment_model_final.ipynb Quantity training
Components 11
Prediction runtime
- Streamlit app
index.py - Quantity model
src/models/xgb_model_qty.pkl - Sales model
src/models/xgb_model_sales.pkl
Data preparation
- Training data
data/training_data.csv - Test data
data/test_data.csv - Feature preparation
src/utils/transformation_util.ipynb - Exploratory analysis
analysis/eda.ipynb
Model development
- Pipeline orchestration
pipelines/run_pipeline.ipynb - Quantity training
src/models/qty_pred_wdepartment_model_final.ipynb - Sales training
src/models/sales_pred_wdepartment_model_final.ipynb
Other components
- Store user
Connections 12
- Store user → Streamlit app: enters inputs
README.md - Streamlit app → Quantity model: loads model
index.py - Streamlit app → Sales model: loads model
index.py - Streamlit app → Store user: shows forecast
index.py - Pipeline orchestration → Exploratory analysis: runs EDA
README.md - Pipeline orchestration → Feature preparation: prepares data
README.md - Pipeline orchestration → Quantity training: builds models
README.md - Pipeline orchestration → Sales training: builds models
README.md - Training data → Feature preparation: prepares records
README.md - Test data → Feature preparation: prepares records
README.md - Quantity training → Quantity model: saves model
README.md - Sales training → Sales model: saves model
README.md