We have a year of time series data which shows building occupancy by the minute.
Is there a way we can leverage AIP to predict a future occupancy time series?
Hi @edery, great question!
When using time series with AIP, consider that large language models may need some help analysing numerical data in a large time series dataset.
The best way to do this in my opinion would be to provide a Typescript function that aggregates over a time series property and serialises that to a string as a tool to AIP Logic.
For example, let’s say I wanted my AIP Logic function to tell the user if building occupancy was increasing, decreasing, higher than average, etc. I could write a Typescript function that would a) see if occupancy was increasing/decreasing, and b) tell me if the occupancy was above/below average and pass this as a string to my logic function for each time bucket I want to look at e.g. 1 hour.
Hope this helps!
JG.