* Transformers: Add smoothing transformer
Added a smoothing transformer to help clean up noisy time series data.
It uses the ASAP algorithm to pick the most important data points while
keeping the overall shape and trends intact.
The transformer always keeps the first and last points so you get the
complete time range. I also added a test for it.
* Change category
Change category from Reformat to CalculateNewFields
* Remove first/last point preservation
* Fix operator recreation
* Simplify ASAP code
Include performance optimization as well
* Refactor interpolateFromSmoothedCurve
Break function into smaller focused functions and lift functions to the
top level
* Add isApplicable Check
Make sure the transformer is applicable for timeseries data
* Add tests for isApplicable check
* UI/UX improvements: Display effective resolution when limited by data points
Show "Effective: X" indicator when resolution is capped by the 2x data
points multiplier. Includes tooltip explaining the limit.
Memoizes calculation to prevent unnecessary recalculation on re-renders.
Example: With 72 data points and resolution set to 150, displays
"Effective: 144" since the limit is 72 x 2 = 144.
Plus added tests
* Improve discoverability by adding tags
* Preserve Original Data
Let's preserve original data as well, makes the UX so much better.
Changed from appending (smoothed) to frame names to use Smoothed frame name. This should match the pattern used by other transformers (e.g,. regression)
Updated tests accordingly
Updated tooltip note
* Add asap tests
Basic functionality:
* returns valid DataPoint objects
* Maintain x-axis ordering
Edge cases:
* Empty array
* single data point
* filter NaN values
* all NaN values
* sort unsorted data
* negative values
* Update dark and light images
* Clear state cache
* Add feature toggle
* Conditionally add new transformation to the registry
* chore: update and regenerate feature toggles
* chore: update yarn.lock
* chore: fix transformers and imports
vesalaakso-oura
a9c2117aa7