Philippe Rivière
Philippe Rivière
The color legends should display the unknown color if it is set, possibly with an unknownLabel that default to "unknown" (or "N/A", "invalid"?) Note that if d3.scale adds scale.high /...
position the text relative to the anchor anchor can be any of "top", "bottom", "left", "right" or combinations such as "top-left" demo: https://observablehq.com/@observablehq/text-facet-alignment-523 closes #523 todo: - [x] create unit...
ref. #80 Todo: - [ ] "grammar of graphics": how to make this work naturally with the other transforms (stackX, groupX) 🌶 - [x] pie mark - [ ] pie...
It would be nice to have an up-to-date overview of all the modules that work and are available in the ml main bundle; also if possible, with working examples (I...
with `main: 'dist/ml.min.js'` in `package.json` I think we could use unpkg.com to load ml directly in the browser (?)
in EigenvalueDecomposition it would be nice to be able to retrieve only the first k eigenvalues & associated vectors, if it could save time and memory. I have no idea...
> RangeError: Array buffer allocation failed this is the error I get if I try to pass more than 30k points to UMAP. (I suppose that DruidJS should accept and...
With the current API, if one wants to project in d=3, one has to know the exact number *n* of optional arguments before specifying 3 as the *n+1*th argument. This...
I took UMAP as an example, but note that many methods crash or return garbage on small datasets. We could probably limit the number of neighbors to *n*-1, etc?
If I understand correctly, the current approach learns and transforms at the same time. As a consequence you can't learn on a subset (train set), then transform the whole dataset....