In this module we will continue building on the analysis and dashboard you built in Module 1. We will explore the Machine Learning (ML) capabilities built into QuickSight such as Narratives, Forecasting, Anomaly Detection and Contribution Analysis.
QuickSight analyzes the time series trends within your data and can spot instances wherein a measure’s overall value or per category value deviated outside the expected value range. Essentially, we are looking for anything that is not normal or ordinary. Surfacing such anomalies can help you identify issues at the onset and take remedial action.
Based on time trends in your data, QuickSight can forecast future values of measures. While turning on forecasts, you also have the option to do backwards forecasting to check if the forecasted values are close to actuals. Say, if you chose to do a backward forecast for 5 months, only data up until 5 months back is fed into forecasting module. Hence, you are able to see 5 months of forecasted data plotted against actual data and then additional forecast for as many periods as you set for forward forecast.
Natural Language Query with Q
Q uses the full power of Machine Learning to interpret questions typed in by user, query underlying datasets and present the response using most apt visual type. It will then let users further customize their experience in form of changing visual types, applying customized filters etc. Q is not covered in this workshop currently. We will be adding a whole section on it when it becomes generally available.