End-to-end demand forecasting and retraining workflow

[#1]

Time series forecasting in DataRobot has a huge suite of tools and approaches to handle highly complex multiseries problems. These include:

• Automatic feature engineering and the creation of lagged variables across multiple data types, as well as training dataset creation.
• Diverse approaches for time series modeling with text data, learning from cross-series interactions, and scaling to hundreds or thousands of series.
• Feature generation from an uploaded calendar of events file specific to your business or use case.
• Automatic backtesting controls for regular and irregular time series.
• Training dataset creation for an irregular series via custom aggregations.
• Segmented modeling, hierarchical clustering for multi-series models, text support, and ensembling.
• Periodicity and stationarity detection and automatic feature list creation with various differencing strategies.
• Cold start modeling on series with limited or no history.
• Insights for models.
• Data and accuracy drift monitoring.
• Automated retraining.

About this Accelerator

This accelerator demonstrates retraining policies with DataRobot MLOps demand forecast deployments.

The dataset consists of 50 series (46 SKUs across 22 stores) over a two year period with varying series history, typical of a business releasing and removing products over time.

DataRobot will be used for the model training, selection, deployment, and making predictions. Snowflake will work as a data source for both training and testing, and as a storage to write predictions back. This workflow, however, applies to any data source, e.g. Redshift, S3, Big Query, Synapse, etc.

What you will learn  

• Ingest the data from Snowflake into AI Catalog within DataRobot
• Run a new DataRobot project
• Deploy the recommended model
• Set up automated retraining
• Define and run a job to make predictions and write them back into Snowflake

Additional Resources

This accelerator is a another instalment of a series on demand forecasting.

• The first accelerator focuses on handling common data and modeling challenges, identifies common pitfalls in real-life time series data, and provides helper functions to scale experimentation.
• The second accelerator provides the building blocks for cold start modeling workflow on series with limited or no history. They can be used as a starting point to create a model deployment for the app.
• The third accelerator is a what-if app that allows users to adjust certain known in advance variable values to see how changes in those factors might affect the forecasted demand.

[#2]

Time series forecasting in DataRobot has a huge suite of tools and approaches to handle highly complex multiseries problems. These include:

• Automatic feature engineering and the creation of lagged variables across multiple data types, as well as training dataset creation.
• Diverse approaches for time series modeling with text data, learning from cross-series interactions, and scaling to hundreds or thousands of series.
• Feature generation from an uploaded calendar of events file specific to your business or use case.
• Automatic backtesting controls for regular and irregular time series.
• Training dataset creation for an irregular series via custom aggregations.
• Segmented modeling, hierarchical clustering for multi-series models, text support, and ensembling.
• Periodicity and stationarity detection and automatic feature list creation with various differencing strategies.
• Cold start modeling on series with limited or no history.
• Insights for models.
• Data and accuracy drift monitoring.
• Automated retraining.