mcorrell

Shortest working example: ``` var test = [{"5.a": 0, "a":1},{"5.a": 2, "a":3}]; dl.summary(test); ```

Added `skew` function that is based on the sample skew functions provided by Excel and suggested by Wikipedia. Old `modeskew` function now points to `npskew` for "non-parametric skew" as a...

Computes the interquartile mean of an array of numbers. If n%4==0, then simply discard the values in Q1 and Q4 and compute the mean of the remaining values. Otherwise, compute...

The wrong skew measure?

4

datalib currently calculates what is supposed to be Pearson's mode skewness, but does so as: `(avg - med) / std` This is [non-parametric skewness](https://en.wikipedia.org/wiki/Nonparametric_skew), rather than: Pearson's [mode skewness](https://en.wikipedia.org/wiki/Skewness#Other_measures_of_skewness) which...

More blending modes for uncertainty

1

Right now uncertainty is encoded by blending to white in LAB space (more uncertainty, more whiteness). We call this mode `usl` (for "uncertainty-saturation-lightness"). We might want more modes. For instance,...

[feature request] More rules for choosing histogram bin sizes

2

Currently `bin` takes an extent, and then some optional combination of explicit or max/min bin size or bin counts. Without those constraints it does a Sturges' rule thing to pick...