uint32 dataframe don't display in `px.bar`

[MarcoGorelli]

Here's an example:

px.bar(
    pd.DataFrame({
        'break_point': [-7, -6., -5, -4, -3],
        'Rain_count': [0, 0, 6, 696, 31335],
    }).astype({'Rain_count': 'uint32'}),
    x='break_point',
    y='Rain_count'
)

If I change uint32 to int32, then it displays fine:

versions:

• pandas: '2.0.3'
• plotly '5.15.0'
• jupyter: 3.6.3

[emilykl]

Oddly enough, this seems to be an issue with the way px.bar() determines orientation! It seems to incorrectly decide that orientation should be "v" for these inputs, which causes the interpretations of x and y to be reversed. If you look very very closely, or zoom in, you can see very thin horizontal bars located at 0, 6, 696, and 31335 along the y-axis.

Forcing orientation="v" results in the expected plot:

Looks like the bug traces to the _is_continuous() function in plotly/express/_core.py, which evaluates to False in the case of Rain_count, causing Plotly Express to decide that Rain_count is the categorical array and should be treated as the bar labels.

It seems the root cause is that dtype.kind for the Rain_count column is u, which is not one of "ifc".

Plausibly we should just change "in "ifc" to in "ifcu" — @MarcoGorelli do you see any issue with that? Looks like that does resolve the issue.

Numpy dtype.kind values: https://numpy.org/doc/stable/reference/generated/numpy.dtype.kind.html

[emilykl]

I think this has the same root cause as #4344