A transform to decimate (sample) data, by filtering the index. Possible strategies: * % sample, not smart but easy to understand * lttb (ref. https://skemman.is/bitstream/1946/15343/3/SS_MSthesis.pdf); adaptive binning. * M4 (as seen in [mosaic](https://observablehq.com/@uwdata/m4-scalable-time-series-visualization)), binning. In practice we probably don't need all the methods; having one by default would be enough. M4 is easy to implement.
A transform to decimate (sample) data, by filtering the index.
Possible strategies:
In practice we probably don't need all the methods; having one by default would be enough. M4 is easy to implement.