Nixtla's Open Source Time Series Ecosystem.

Nixtla Community

20230907_220940.jpg

Using hierarchicalforecast, with MiddleOut (forecast proportions) reconciliation, it happened that for a certain point the second-to-last level has a forecasted value of 4 and the last (lowest) level, for the same timestep, has a forecasted value of over 7000 (I expect it to be equal or lower than 4). Could it be related to some division by a value close to zero that is happening somewhere?

The problem was caused by some very small negative values close to zero (from the chart they looked like zero because they were very small). By clipping to zero as the minimum value, the problem seems to have been fixed.