如何绘制时间序列，其中x轴是matplotlib中的datetime.time对象？

Question

I'm trying to graph timeseries data beginning from 9pm to 6pm the next day. Here is my failed attempt.

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from datetime import time    

a=np.array([35,25,24,25,27,28,30,35])
df=pd.DataFrame(index=pd.date_range("00:00", "23:00", freq="3H").time,data={'column1':a})

          column1
00:00:00       35
03:00:00       25
06:00:00       24
09:00:00       25
12:00:00       27
15:00:00       28
18:00:00       30
21:00:00       35

reindexing data to go from 21:00 to 18:00. perhaps there is a better way of achieving this part too, but this works.

df=df.reindex(np.concatenate([df.loc[time(21,00):].index,df.loc[:time(21,00)].index[:-1]]))

          column1
21:00:00       35
00:00:00       35
03:00:00       25
06:00:00       24
09:00:00       25
12:00:00       27
15:00:00       28
18:00:00       30

plt.plot(df.index,df['column1'])

the x-axis does not seem to match df.index. also the axis begins at 00:00 not 21:00. does anyone know a solution that doesn't involve using string labels for the x-axis?

Answer 1

A simple way to do that is to plot the data without expliciting the x axis and the to change the labels. The problem is that this will only works if the time between the data is constant. I know you said you didn't want to use string labels, so I don't know of this solution will be what you want.

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from datetime import time    

a=np.array([35,25,24,25,27,28,30,35])

df=pd.DataFrame(index=pd.date_range("00:00", "23:00",freq="3H").time,data={'column1':a})

df=df.reindex(np.concatenate([df.loc[time(21,00):].index,df.loc[:time(21,00)].index[:-1]]))

# Now we create the figure and the axes
fig,axes=plt.subplots()
axes.plot(df['column1'])  # In the x axis will appear 0,1,2,3...
axes.set_xticklabels(df.index)  # now we change the labels of the xaxis

plt.show()

This should do the trick and will plot what you want.