Auto Predict Not Accounting for Recurring Missing Months Trend

Summary:

Auto Predict forecasting negative values for recurring missing months and ignoring missing-month trends or events in historical data

Content (please ensure you mask any confidential information):

We are using Auto Predict with 24 months of historical data. In the historical data, there is a recurring pattern of missing volume from July to Oct. However, Auto Predict does not seem to treat this missing pattern as part of the trend and instead generates negative values for July–Oct in the forecast.

We have defined events using Repeat or Skip, but this does not resolve the issue. We also enabled Populate Missing Values