Investigate using LTTB for maxDataPoints #154
Comments
|
Can you provide more details about that? |
|
Yeah, but I'd like to get more information about what we were expecting from that :) |
|
The idea is that it provides a more visually representative view of the data than other downsampling algorithms. The original paper is at http://skemman.is/stream/get/1946/15343/37285/3/SS_MSthesis.pdf . It seems to have been pretty well adopted in the time since, and you can find some convincing visual results if you search the web. |
|
Two things I ran into when looking into this: 1) figuring out the API and where exactly in carbonapi this would get implemented and 2) LTTB doesn't return evently spaced (in time) points, which means it doesn't follow the start/stop/step pattern (which makes it tricky to fit into the responses we have). If grafana is happy with this, then we're good. |
…functions (#154) * Take specificed bootstrapInterval into consideration when adjusting holtWinters start times * Adjust holtWintersConfidenceArea from * Add tests for Metrics() function adjusting holtWinters from time
…functions (#154) * Take specificed bootstrapInterval into consideration when adjusting holtWinters start times * Adjust holtWintersConfidenceArea from * Add tests for Metrics() function adjusting holtWinters from time
and others
Maybe
consolidateBy=lttb?https://github.com/dgryski/go-lttb
The text was updated successfully, but these errors were encountered: