Option to center labels on bins about physt HOT 5 CLOSED

Seems really like "histogram as bar plot kind of thing", because from the distribution-describing point of view, a bin is always a range (1.0 to 2.0, or 1.5 to 2.5).

Anyway, some approximation of what you mention should be possible (ticks="center" kind of parameter to force this). Or perhaps a generic "category" histogram solution (which I want to include) would be applicable? Let me think about it.

from physt.

Understood.

I often have quantized data with few and integer values, but nevertheless want to plot the count of. If the MPL bar plot API was not so terrible then I wouldn't file an issue here ;-)

I would not necessarily consider this as categorical data.

from physt.

Ok, I included (to support your use case):

• integer binning (bins centered around integers)
• ticks="center" argument for plot method

Example:

# Sum of two dice (should be triangle, right?)
dice = np.floor(np.random.rand(10000) * 6) + np.floor(np.random.rand(10000) * 6) + 2
histogram(dice, "integer").plot(ticks="center", density=True);

See https://github.com/janpipek/physt/blob/731d0736bb3667d455c348e9d76676df97b8e1aa/doc/Binning.ipynb for the representation :-)

Is that what you wanted?

from physt.

P.S. Currently supported just for matplotlib. I like bokeh but... later :-)

from physt.

Yip, that meets my use-case. Thanks a lot!

FWIW I think supporting two plotting backends is just making extra work for yourself - I think for better or worse MPL is all you need, as ugly as it can be ;-)

from physt.