BUG: Wrong legend when combining multiple plots on the same axes

[david-zwicker]

The code

import matplotlib.pyplot as plt
import pandas as pd

df = pd.DataFrame([{'x': 1, 'a': 1, 'b': 1}, {'x': 2, 'a': 2, 'b': 3}])

ax = plt.gca()
df.plot('x', 'a', yerr=0.1, ax=ax, label='with errors')
df.plot('x', 'b', ax=ax, label='without errors')

results in the following image on my platform:

Note that the labels in the legend are swapped!
I obviously expected the legend to show the correct labels.

INSTALLED VERSIONS
------------------
commit           : 9d598a5e1eee26df95b3910e3f2934890d062caa
python           : 3.8.7.final.0
python-bits      : 64
OS               : Darwin
OS-release       : 19.6.0
Version          : Darwin Kernel Version 19.6.0: Tue Nov 10 00:10:30 PST 2020; root:xnu-6153.141.10~1/RELEASE_X86_64
machine          : x86_64
processor        : i386
byteorder        : little
LC_ALL           : None
LANG             : en_US.UTF-8
LOCALE           : en_US.UTF-8
pandas           : 1.2.1

numpy            : 1.19.5

pytz             : 2020.5

dateutil         : 2.8.1

pip              : 21.0

setuptools       : 51.3.3

Cython           : 0.29.21

pytest           : 6.2.2

hypothesis       : None

sphinx           : 3.4.3

blosc            : None

feather          : None

xlsxwriter       : None

lxml.etree       : 4.4.1

html5lib         : 1.1

pymysql          : None

psycopg2         : None

jinja2           : 2.11.2

IPython          : 7.19.0

pandas_datareader: None

bs4              : None

bottleneck       : None

fsspec           : None

fastparquet      : None

gcsfs            : None

matplotlib       : 3.3.4

numexpr          : None

odfpy            : None

openpyxl         : None

pandas_gbq       : None

pyarrow          : None

pyxlsb           : None

s3fs             : None

scipy            : 1.6.0

sqlalchemy       : None

tables           : None

tabulate         : 0.8.7

xarray           : None

xlrd             : None

xlwt             : None

numba            : 0.52.0

[robertzee]

Same issue here.

[attack68]

if you try:

df.plot('x', 'a', yerr=0.1, ax=ax, label='one red', color='r')
df.plot('x', 'b', ax=ax, label='two green', color='g')
df.plot('b', 'a', ax=ax, label='three blue', color='b')
df.plot('b', 'a', ax=ax, label='four yellow', color='y')

The legend shows the text in the right order but the colours have been permuted backwards by one:

<green> one red
<blue> two green
<yellow> three blue
<red> four yellow

If you remove yerr=0.1 the problem disappears.
If you add yerr=0.1 to all methods the problem disappears.

If you have 2 methods with yerr=0.1 then the color permutation shifts by 2 instead of one, for 3 the permutation is 3.

[nmay231]

I assume the culprit is somewhere around this section of code, but I don't understand how the kwargs are shifted across calls to df.plot().

Unless there is something funky with how the matplotlib wrapper applies keyword arguments on shared plots (in which case, maybe this function is messed up? 🤷).

Also note that removing yerr from @attack68's first df.plot() and adding it to the last one switches only the last two plot colors.

ax = plt.gca()
# Only blue and yellow are switched now; red and green are fine
df.plot('x', 'a', ax=ax, label='one red', color='r')
df.plot('x', 'b', ax=ax, label='two green', color='g')
df.plot('b', 'a', ax=ax, label='three blue', color='b')
df.plot('b', 'a', yerr=0.1, ax=ax, label='four yellow', color='y')

[MarcoGorelli]

Thanks @david-zwicker for the report - indeed, I can confirm this reproduces on master, will look into it

[MarcoGorelli]

Just tried this out with matplotlib directly, and it displays fine, so the bug is probably to do with pandas

---

this is related to #40044

[MarcoGorelli]

Update: since #40777 , the colours in the legend are now correct. However, due to an open issue in matplotlib, marker info is not yet preserved in the legend, so let's keep this open until then