Grid=True option in plot functions overrides user settings

[mangecoeur]

The plotting functions on DataFrames always have grid=True. This overrides user settings in matplotlibrc.

If you have to created a desired default plot style with the rc file, you want every plot to show with this style unless otherwise specified.
Because pandas overrides this setting you have to manually set the grid option every time.

A better behaviour would be for Pandas not to override any plot styles by default.

[ghost]

Marking as 0.12, though I'm not sure that will happen unless someone steps up
to overhaul the way settings are handled.
If you're comfortable working on this, a PR would be welcome.

[goyodiaz]

I want to work on this but I did not find any tests of plotting stuff. Should we add a test_plotting module?

[jreback]

pandas/tseries/tests/test_plotting.py

[cpcloud]

Also pandas/tests/test_graphics.py for plotting not unrelated to time series.

[jseabold]

Came here to file this same issue so +1.

[WillAyd]

While not all plotting functions have grid=True it is rather inconsistent in the code base as of v0.23 and can use some simple cleanup

[jeffgeee]

take

[bdrum]

Hi everyone! I would like to take this issue, but I can't reproduce it for fresh built version.
From my side the behavior is correct:

import numpy as np
import pandas as pd
from matplotlib import pyplot as plt
import matplotlib as mpl

print(pd.__version__)
print("mpl.rcParams['axes.grid']", mpl.rcParams['axes.grid'])
df = pd.DataFrame(columns=["A", "B"], data=np.random.rand(100,2))
df.plot(grid=True)
print("mpl.rcParams['axes.grid']", mpl.rcParams['axes.grid'])
f = plt.figure()
_ = f.gca().plot(np.random.rand(100))

# output
1.4.0.dev0+921.gef9416c51c
mpl.rcParams['axes.grid'] False
mpl.rcParams['axes.grid'] False

As I can see it was appeared in 0.12 first time, but looks like now every thing is fine. I also have tried set rc.axes.grid as True in the file and then run the same script but with explicit grid=False argument and without it. Works correct: when grid=False it doesn't show grid despite of the rc params and in case of grid doesn't specified it has the same behavior as in file specified i.e. grid will be shown.

[bdrum]

take

[Dr-Irv]

@bdrum If I understand correctly, the issue is that if you have a matplotlibrc file (https://matplotlib.org/stable/tutorials/introductory/customizing.html ), and that file specifies axes.grid=False, then you still get the grid, and it is suggested here that pandas should not override what is specified in the file. Your test above doesn't check that.

[bdrum]

@Dr-Irv thanks for clarification.
Yes, as it shown you will get the grid in case of explicit setting of grid parameter, but in case of you miss it, the behavior will synced with rc settings:

import numpy as np
import pandas as pd
import matplotlib as mpl
print(pd.__version__)
print("mpl.rcParams['axes.grid']", mpl.rcParams['axes.grid'])
df = pd.DataFrame(columns=["A", "B"], data=np.random.rand(100,2))
df.plot()

# output 
1.4.0.dev0+921.gef9416c51c
mpl.rcParams['axes.grid'] True

and revert back to grid false in rc file:

import numpy as np
import pandas as pd
import matplotlib as mpl
print(pd.__version__)
print("mpl.rcParams['axes.grid']", mpl.rcParams['axes.grid'])
df = pd.DataFrame(columns=["A", "B"], data=np.random.rand(100,2))
df.plot()
# output 
1.4.0.dev0+921.gef9416c51c
mpl.rcParams['axes.grid'] False

According to discussion above, I can try to add some checks in tests

[Dr-Irv]

According to discussion above, I can try to add some checks in tests

If you can create a test that verifies this in a PR, that would be great.

[bdrum]

@Dr-Irv
As I can see this issue is a duplicate of #9792. The last one was closed and tested here.

[Dr-Irv]

@bdrum Thanks. I will close this as a result of your discovery.