SKU-Level Demand Forecasting
8% accuracy uplift using SARIMA/VARMAX for 1,000+ automotive SKUs.
PythonSARIMAVARMAXForecastingAutomotiveMachine Learning
About the Project
This project involved developing Python-based forecasting models using SARIMA and VARMAX algorithms for 1,000+ automotive SKUs in the spare parts network.
The challenge was to improve forecasting accuracy for a highly complex SKU portfolio with varying demand patterns, seasonality, and interdependencies. Traditional forecasting methods were proving inadequate.
By implementing advanced time-series algorithms and incorporating external variables, we achieved an 8% accuracy uplift in SKU-level forecasting, enabling better inventory planning and significantly reducing stockouts.
Key Features
Python-based forecasting implementation
SARIMA algorithm for seasonal patterns
VARMAX for multivariate analysis
1,000+ SKU coverage
External variable integration
Inventory planning optimization
Challenges & Learnings
Managing highly complex SKU portfolio
Handling varying demand patterns and seasonality
Integrating forecasts into operational planning