v1.0.0

First tagged release of RetailSense Lite: a retail analytics dashboard that forecasts demand per product with a Prophet, XGBoost and LightGBM ensemble, flags unusual sales, raises inventory alerts and simulates pricing changes, all in a Streamlit app.

Changes in this release

• Fixed a crash on current pandas: 12 calls to fillna(method="ffill"|"bfill"), removed in pandas 3, now use .ffill() and .bfill(). Anomaly detection raised a TypeError on a fresh install before this
• Added unit tests for price optimisation, price-elasticity estimation and anomaly detection, using synthetic data so no dataset is needed
• Added CI that runs the tests on every push
• Added a README with features, the forecasting approach, setup and the expected dataset schema

Known limits

• The dataset is not in the repository. The dashboard needs data/processed/data_with_all_features.csv, described in the README
• The tests cover the business-logic helpers, not the Streamlit UI or the full forecasting pipeline
• The one documented accuracy figure (ensemble R-squared about 0.77, one product, 13-week horizon) comes from the project's own test script and is not a cross-product benchmark