[FIXED] pandas plot ignores index

Issue

I have a pandas Series with integer running hours as index, where the hours do not start at midnight but rather on a random hour (8 in the example below).

hour = range(8, 24) + range(0, 8)
values = pd.Series(range(24), index=hour)

I wish to plot a 24-hours graph of the values, but I want the x-axis to show the "raw" hour values and not re-arrange the index.

ax = values.plot()
ax.axis([0, 23, 0, 23])

Any suggestions?

Solution

One option is indeed (as you said yourself in the comments) to use a non-numeric index.

import pandas as pd
import matplotlib.pyplot as plt

hour = range(8, 24) + range(0, 8)
values = pd.Series(range(24), index=list(map(str,hour)))

ax = values.plot()

plt.show()

Since this will automatically choose some labels and there is no way to control them, it might not be optimal.

A better solution is probably to use a different index, namely one that is numeric and continuously increasing (in this case from 8 to 31) and then modify the ticklabels as the modulo of 24.

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.ticker

hour = range(8, 24) + range(0, 8)
values = pd.Series(range(24), index=hour)
values.index = values.index[0]+np.arange(0,len(values))

ax = values.plot()
func = lambda x,pos: "{:g}".format(x%24)
ax.xaxis.set_major_formatter(matplotlib.ticker.FuncFormatter(func))

plt.show()

This would then also allow to set the locations to multiples of some number (e.g.6) which makes sense for hours in a day.

ax.xaxis.set_major_locator(matplotlib.ticker.MultipleLocator(6))

Finally you could of course opt for using real datetimes.

import pandas as pd
import matplotlib.pyplot as plt
from datetime import datetime

hour = range(8, 24) + range(0, 8)
values = pd.Series(range(24), index=hour)

todate = lambda i: datetime(2017,11,8+(values.index[0]+i)//24,(values.index[0]+i)%24)
values.index = [todate(i) for i in range(len(values))]

ax = values.plot()

plt.show()

Answered By - ImportanceOfBeingErnest