I’m using the AWS Cloud Intelligence Dashboard (CID) and I’m trying to understand the ‘Forecast’ option within the main visuals. I can see the dashboard shows a projection of future costs, but I’m not sure how this prediction is being generated.
My question is:
Is the dashboard pulling this data from the Amazon Forecast service in the background, or does it use a built-in QuickSight forecasting function based on the aggregated CUR data?
I’m asking because I’d like to understand the underlying model and its capabilities. I’m exploring ways to create more advanced forecasts that incorporate our specific business events and resource tags, and I’m trying to figure out if I can customize the existing CID forecast or if I need to build a separate predictive model using a service like SageMaker Canvas.
First of all Welcome to the QuickSight community !
The forecasting in AWS Cloud Intelligence Dashboards uses QuickSight’s built-in forecasting capabilities, not Amazon Forecast service.
QuickSight uses its own built-in forecasting algorithm called BIFOCAL rather than pulling data from a separate Amazon Forecast service. Reference here
This algorithm utilizes two Random Cut Forests (RCFs) in a CALibrated BI-FOrest architecture, where the first RCF filters out anomalies and provides an initial forecast, which is then corrected by the second RCF.
The QuickSight forecasting approach is designed to be robust and requires fewer parameters than other widely available algorithms, making it useful without human adjustment for larger numbers of time series data points.