hvplot.hist(..., logy=True) is broken

[stuarteberg]

The hvplot.hist() function cannot be used with the logy option, even when all bins are populated and therefore ought to have non-infinite log values.

ALL software version info

relevant package versions

``` $ conda list | grep -E 'python|numpy|^pandas|bokeh|holo|hvplot|jupyter' bokeh 3.4.1 pyhd8ed1ab_0 conda-forge brotli-python 1.1.0 py312h30efb56_1 conda-forge holoviews 1.18.3 pyhd8ed1ab_0 conda-forge hvplot 0.10.0 pyhd8ed1ab_0 conda-forge ipython 8.24.0 pyh707e725_0 conda-forge jupyter-lsp 2.2.5 pyhd8ed1ab_0 conda-forge jupyter_bokeh 4.0.4 pyhd8ed1ab_0 conda-forge jupyter_client 8.6.2 pyhd8ed1ab_0 conda-forge jupyter_core 5.7.2 py312h7900ff3_0 conda-forge jupyter_events 0.10.0 pyhd8ed1ab_0 conda-forge jupyter_server 2.14.0 pyhd8ed1ab_0 conda-forge jupyter_server_terminals 0.5.3 pyhd8ed1ab_0 conda-forge jupyterlab 4.2.1 pyhd8ed1ab_0 conda-forge jupyterlab_pygments 0.3.0 pyhd8ed1ab_1 conda-forge jupyterlab_server 2.27.2 pyhd8ed1ab_0 conda-forge jupyterlab_widgets 3.0.11 pyhd8ed1ab_0 conda-forge libboost-python 1.82.0 py312hfb10629_6 conda-forge libboost-python-devel 1.82.0 py312h8da182e_6 conda-forge msgpack-python 1.0.8 py312h2492b07_0 conda-forge neuprint-python 0.4.26+8.gd487a3c dev_0 numpy 1.26.4 py312heda63a1_0 conda-forge openmesh-python 1.2.1 py312hfb8ada1_4 flyem-forge pandas 2.2.2 py312h1d6d2e6_1 conda-forge python 3.12.3 hab00c5b_0_cpython conda-forge python-blosc 1.11.1 py312hfb8ada1_0 conda-forge python-dateutil 2.9.0 pyhd8ed1ab_0 conda-forge python-fastjsonschema 2.19.1 pyhd8ed1ab_0 conda-forge python-json-logger 2.0.7 pyhd8ed1ab_0 conda-forge python-snappy 0.7.1 pyh48db8ab_1 conda-forge python-tzdata 2024.1 pyhd8ed1ab_0 conda-forge python_abi 3.12 4_cp312 conda-forge types-python-dateutil 2.9.0.20240316 pyhd8ed1ab_0 conda-forge ```

conda info

``` $ conda info active environment : flyem-312 active env location : /groups/flyem/proj/cluster/miniforge/envs/flyem-312 shell level : 1 user config file : /groups/flyem/home/bergs/.condarc populated config files : /groups/flyem/proj/cluster/miniforge/.condarc conda version : 24.5.0 conda-build version : 24.3.0 python version : 3.8.13.final.0 solver : libmamba (default) virtual packages : __archspec=1=sapphirerapids __conda=24.5.0=0 __glibc=2.34=0 __linux=5.14.0=0 __unix=0=0 base environment : /groups/flyem/proj/cluster/miniforge (writable) conda av data dir : /groups/flyem/proj/cluster/miniforge/etc/conda conda av metadata url : None channel URLs : https://conda.anaconda.org/flyem-forge/linux-64 https://conda.anaconda.org/flyem-forge/noarch https://conda.anaconda.org/conda-forge/linux-64 https://conda.anaconda.org/conda-forge/noarch package cache : /groups/flyem/proj/cluster/miniforge/pkgs /groups/flyem/home/bergs/.conda/pkgs envs directories : /groups/flyem/proj/cluster/miniforge/envs /groups/flyem/home/bergs/.conda/envs platform : linux-64 user-agent : conda/24.5.0 requests/2.28.1 CPython/3.8.13 Linux/5.14.0-362.8.1.el9_3.x86_64 rocky/9.3 glibc/2.34 solver/libmamba conda-libmamba-solver/24.1.0 libmambapy/1.5.8 UID:GID : 50328:93098 netrc file : None offline mode : False ```

Description of expected behavior and the observed behavior

When using logy=True, no bars are shown, and a warning is given. (See below.)

Complete, minimal, self-contained example code that reproduces the issue

import pandas as pd
import hvplot.pandas

data = pd.Series([1, 1, 2, 2, 2, 3, 3, 3])
data.hvplot.hist(bins=[0,2,4], logy=True)

Stack traceback and/or browser JavaScript console output

WARNING:param.HistogramPlot08409: Logarithmic axis range encountered value less than or equal to zero, please supply explicit lower bound to override default of 0.010.

Screenshots or screencasts of the bug in action

[hoxbro]

Thank you for the issue.

This is a limitation in Bokeh, https://github.com/bokeh/bokeh/issues/6536 and is also tracked in HoloViews here.

[stuarteberg]

@hoxbro Thanks for the response. I'm happy to close this if it's truly a duplicate of the two issues you linked to.

I'm confused, though: Those issues pertain to data with at least one empty bin. In my case, there are no empty bins, so the minimum y-value should be non-negative, even after taking the log.

In [12]: data = [1, 1, 2, 2, 2, 3, 3, 3]
    ...: y = np.histogram(data, bins=[1,2,4])[0]
    ...: np.log(y)
Out[12]: array([0.69314718, 1.79175947])

[hoxbro]

The zero referred to is the bottom of the rectangles used for the histogram bars, not the empty space between bars.

[stuarteberg]

Ah, I see. So it's not the y-values that matter... that would be the top of the bars. But every bar has the same bottom position, and it's 0 for all of them.

Thanks for the clarification. Closing this.

[ampilnick]

Please don't close this issue until it will not fixed! Now hvplot.hist(..., logy=True) is still broken.

[maximlt]

Hi @ampilnick, I'm -1 on re-opening this issue as it is already tracked in https://github.com/holoviz/holoviews/issues/2591 and https://github.com/bokeh/bokeh/issues/6536. I have added the hvplot label to https://github.com/holoviz/holoviews/issues/2591 to let the maintainers know that this also affects hvPlot.