dataset_description.md

<!-- Prepared by Molla Samser (RSK World) | hello@rskworld.in | support@rskworld.in | +91 93305 39277 | Nutanhat, Mongolkote, Purba Burdwan, West Bengal, India 713147 | https://rskworld.in -->

# Dataset Description

## Overview
This dataset is designed for sales forecasting and demand planning. It combines transactional sales records, product attributes, customer segments, promotions, and time-aware features to enable classical time series models and feature-based regressors.

## Schema (sample_sales.csv)
- `date` — ISO date of sale.
- `store_id` — store identifier.
- `product_id` — product identifier.
- `product_category` — category label for grouping.
- `price` — unit price on the transaction date.
- `units_sold` — quantity sold.
- `revenue` — `price * units_sold`.
- `customer_segment` — segment tag such as retail/online/B2B.
- `promotion` — promo descriptor active on date.
- `stock_level` — on-hand inventory near time of sale.
- `region` — store region for geographic patterns.

## Targets
- Primary: `units_sold`
- Secondary: `revenue`

## Recommended Features
- Date parts: day, week, month, quarter, year, day-of-year.
- Seasonal Fourier terms for yearly cycles (sin/cos over day-of-year).
- Rolling statistics: 4-week and 12-week means/sums for `units_sold` and `revenue`.
- Lag features: 7/14/28-day (per-week) lags on `units_sold` and `revenue`.
- Categorical encodings: product_category, region, customer_segment.
- Price/promo indicators: price deltas, promo flags, discount depth where available.
- Inventory coverage: `stock_level` divided by average daily demand.
- Elasticity probes: scatter/partial dependence of price vs `units_sold`.

## Usage Tips
- Keep time order when splitting train/validation (e.g., 80/20 last split).
- Check stationarity and seasonality before picking models.
- Track holiday and campaign calendars in `promotion`.
- Scale/normalize only on training splits; reuse scalers for validation/test.

## Deliverables to Prepare
- Full dataset CSVs in `data/`.
- Packaged archive at `sales-forecasting.zip`.
- Thumbnail/preview image at `sales-forecasting.png`.