Building a successful inventory planning solution with Amazon Forecast
summary
Amazon Forecast claims up to 50% more accurate forecasts and adds features like a Weather Index and holiday regressors; More Retail tests DeepAR+ models and moves from many per-SKU ARIMA models to two global models fed from Redshift via S3 and EC2 Python jobs orchestrated by Apache Airflow. The production pipeline writes forecasts back into S3, auto-generates indents into Oracle ERP, and yields concrete results: accuracy jumps from 24% to ~70%, 30% waste reduction, and a 25% profitability bump.
AI-written summary · about
chapters (11)
- 0:00Introduction to Amazon Forecast
- 1:05History of forecasting at Amazon
- 4:27Key steps for inventory planning
- 8:04Using the Amazon Forecast console
- 11:09Training models and optimizing accuracy
- 16:23Generating forecasts with Amazon Forecast
- 17:26More Retail's success with Amazon Forecast
- 21:55More Retail's forecasting journey
- 27:45Deployment architecture at More Retail
- 31:23Keys to success with Amazon Forecast
- 33:53Q&A and resources
related talks
25mTransform your business with data-driven insights
28mDemand Forecasting: Accuracy, Stability, Automation
16mTransforming Inventory Forecasting with Production ML
23mMachine Learning with a Deep Dive into Automated Forecasting System
28mHow Human Expertise and AI drive Scalable Forecasting Success at
30mAI in Retail - Advancing Promo Demand Forecasting
40mMulti-Series Forecasting at Scale with StatsForecast
31mFrom stockouts to happy customers: Proven solutions for time series forecasting in retail