title	emoji	colorFrom	colorTo	sdk	app_port	tags	pinned	short_description	license
TimeSeriesForecasting	📈	blue	green	docker	8501	streamlit	false	LightGBM-based time series forecasting app that predicts...	mit

📈 Sales Forecast (LightGBM)

This project is a Streamlit forecasting app that predicts num_sold using a trained LightGBM model with:

• calendar features (year/month/week/dayofweek/dayofyear, weekend)
• lag features (lag_364, lag_365, lag_371)
• categorical inputs (country/store/product) via saved encoders

Goal: Provide a lightweight, fast baseline forecasting solution with strong feature engineering and a deployable app.

🔗 Live Demo & Code

🤗 HuggingFace Space: [https://huggingface.co/spaces/EnYa32/TimeSeriesForecasting]

💻 GitHub Repository: [https://github.com/EnYa32/TimeSeriesForecasting]

📓 Kaggle-Wettbewerb: [https://www.kaggle.com/code/enesyama/timeseries-forecasting]

What it does

• Takes calendar features (year/month/week/dayofweek/dayofyear, weekend)
• Uses lag features (lag_364, lag_365, lag_371)
• Uses categorical inputs (country/store/product) via saved encoders
• Outputs a num_sold prediction

📊 Modelldiagnose

Restwerte im Zeitverlauf

Istwert vs. Vorhersage

Restverteilung

Wichtigkeit der Merkmale

Files required (put in the repo root)

• app.py
• lgbm_model.pkl
• feature_names.pkl (list of feature names in correct order)
• encoders.pkl (dict of LabelEncoders for country, store, product)
• fill_map.pkl (optional: medians for numeric feature filling)

How to save artifacts in your notebook (training side)

import joblib

joblib.dump(model_lgb, 'lgbm_model.pkl')
joblib.dump(FEATURES, 'feature_names.pkl')
joblib.dump(encoders, 'encoders.pkl')

# optional numeric medians for filling missing
num_cols = [c for c in FEATURES if c not in ['country', 'store', 'product']]
fill_map = train_fe[num_cols].median().to_dict()
joblib.dump(fill_map, 'fill_map.pkl')