Retail Demand Forecast

Time-series demand forecasting for retail/SKU level. Baseline model (LSTM or XGBoost) trained on historical sales to predict future demand. Suitable for inventory and planning.

Overview

Predicts next-period (e.g. daily or weekly) demand from past history. Can be extended to multiple SKUs or stores; the baseline is single-series for simplicity.

Data

CSV with columns such as date and demand (or sales). A sample is in data/demand_sample.csv. If --data_path is missing or the file does not exist, the script uses a synthetic series so you can run without your own data.

Usage

pip install -r requirements.txt
python train.py

Use --data_path to point to your CSV. Model checkpoints are written to ./checkpoints/.

Files

model.py — LSTM forecaster; train.py — training (synthetic fallback if no CSV).
data/demand_sample.csv — example CSV schema.

Model

LSTM (PyTorch) on sliding windows; saves to ./checkpoints/demand_forecast.pt.

Limitations / future work

Single series baseline; multi-SKU would need a global model or per-SKU training.
Optional: upload a small public dataset to Hugging Face Datasets for reproducibility.

Author

Alireza Aminzadeh

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support