'Negative ARIMA forecast possible
Is it possible to have negative forecast in ARIMA (specifically BigQueryML ARIMA_PLUS) when training data contains only 0 or greater than 0 values ? Assume no missing timestamps.
Also if there can be negative forecasts what are the ways to reduce such forecasts. I can only think of post-processing step which is making negative forecasts as 0. But are there other ways ?
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|