tsdownsample

tsdownsample copied to clipboard

This GitHub aims to improve the computational performance of the `MinMaxLTTB`. To clarify, MinMaxLTTB currently utilizes the `x` data in the minmax prefetching step. This causes to create equidistant bins,...

jonasvdd

Different results between different implementations

3

I have been playing around with your implementation of lttb and trying to compare it to other implementations (numpy and numba) and have been impressed by the speed.  But...

hoxbro

*unsure if this will improve the performance, but seems worth exploring* :eyes: I believe we could use smallvec to write the output (usize array) of the downsampling methdos to. Currently...

jvdd

Is a constant sampling rate required?

2

In your _Usage_ window, it says that a constant sampling rate is assumed. Is that generally the case, or just in the example? Under _Limitations and Assumptions_ nothing is mentioned,...

lcs-crr

Is there a benchmark comparing the algorithms with "classic" low-pass filter + decimation from control systems?

lcs-crr

chore: update ruff command and specs

1

jvdd

:muscle: add FPCS downsampling algorithm

1

This PR ports the Feature-Preserving Compensated Sampling algorithm to tsdownsample https://ieeexplore.ieee.org/document/10669793 DONE: - [x] tests for fpcs - [x] (compare with vanilla python / rust implementation) - [x] Nan handling...

jonasvdd

fix: prune nan idxs in output when omitting nans

1

closes #73 This PR - removes NaNn idxs when non NaN downsamplers (i.e., policy = omit NaN - such as `MinMaxDownsampler`) - nothing is changed for NaN downsamplers (i.e., policy...

jvdd

Add nightly feature

2

Currently downsample_rs cannot be built with stable compiler, because `nightly_simd` is enabled by default for `argminmax`. I've added `nightly` feature (which is enabled by default) which activates `nightly_simd`, so this...

leviska

New release 0.1.5

2

my1e5