01
Business needs
A Fortune 500 global groceries and merchandise retailer wanted to add forecasting capabilities to its centralized stock management system by leveraging big data and ML practices. The global retailed wanted to migrate its legacy infrastructure to a Spark-based architecture to advance its analytical initiatives and achieve the following:
- Provide short and long-range forecasting for all store products
- Enhance the overall forecasting accuracy
- Improve revenue by leveraging advanced analytics
- Improve sales by understanding user’s buying preferences and behavior
50% improvement in overall warehousing utilization and management
02
Solution
The Impetus team supported the global retailer to improve its overall demand-supply chain with statistical algorithm leveraging forecasting model based on a big data platform using distributed frameworks such as Hadoop and Spark. It includes the following highlights:
- Enabled short-range forecasting (SRF) and long-range forecasting (LRF) by implementing algorithms that provide more accurate estimates while minimizing noise for the forecasting of all store products (from 28 days to up to 52 weeks)
- Developed a solution that optimizes forecasting jobs by taking into consideration various interventions like weather, promotions, and more
- Implemented dedicated solutions to overcome file-related problems and avoid the overloading of clusters
- Implemented an added feature for enhanced Estimated Daily Sales (EDS) to provide forecasting for online stores
- Added enhancements for new line items, count line items, and various features in interventions to streamline the retail demand-supply process
- Enhanced demand forecasting related modules to improve operational efficiencies
Optimized various jobs in the demand-supply model, improving their processing time while reducing resource utilization by up to 2x
03
Impact
An advanced forecasting model enabled the global retailer to:
- Enhance the downstream system by providing accurate forecasts for all store products
- Improve the overall warehousing utilization and management by up to 50%
- Optimize various jobs in the demand-supply model, improving their processing time while reducing resource utilization by up to 2x
- Ensure that all jobs were completed within their SLAs
- Significantly improve the overall customer rating
