Thin out datasource by aggregation over time period

[kralan]

Dear maintainer, this is only a wishlist item.

If possible, please add functionality to clean a data source representing a time series by aggregating over a time period.

Use case: I have data sources of environmental data I add via the API. I sample the data every 10 minutes. This much detail is interesting only a few days back. After that, I'd like to keep the data for long-term statistics, but I don't need that much detail any more.

Feature request: I'd like to have a cleanup filter like: For data older than a cutoff point in time (eg 7 days ago), keep only one data point per period (eg of 3 hours) by aggregating all data that fall into one such period by an aggregation function (eg average).

I can solve this by writing a query that operates directly on the database, but I thought this might be a good enhancement for the analytics app and it can be useful for any time-series data. Maybe this can be even further generalized to have aggregation intervals other than time.

[Rello]

Hi,
yes, this is a valid idea. in general to have an aggregation of values across time.
I am not sure at the solution yet. One idea might be a datasource on repots itself. e.g. have the data in detail and then load &delete the aggregated data within Analytics.

This would make it flexible in a way that the logic does not need to be implemented in various data source logic parts

...lets see