在pandas / matplotlib中格式化时间轴x轴

Question

I would like to show each month abbreviation, as well as the year on the year.

I am quite close. The issue I am currently having is that the years are incorrect. I have figured out that this is a issue between numpy.datetime64 (the datetime index is in this format), and python datetime which is used the 1970 epoch. The two years shown on the chart should be 2017 and 2018 but they show 48 and 49.

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

from matplotlib.dates import MonthLocator, WeekdayLocator, DateFormatter, YearLocator

indx = pd.date_range('2017-04-01', '2019-01-01')
s = pd.Series(np.random.randn(len(indx)), index=indx)

df = pd.DataFrame(s)

ax = df.plot()
months = MonthLocator(range(1, 13), bymonthday=1, interval=1)
monthsFmt = DateFormatter("%b")
years = YearLocator(1, month=4, day=1)
yrsFmt = DateFormatter("\n %y")

ax.xaxis.set_major_locator(years)
ax.xaxis.set_major_formatter(yrsFmt)


ax.xaxis.set_minor_locator(months)
ax.xaxis.set_minor_formatter(monthsFmt)

plt.show()

How do I show the right years here?

Answer 1

After some playing around it seems to work if you specify the axis and then plot onto that (rather than call the pandas plot function).

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

from matplotlib.dates import MonthLocator, WeekdayLocator, DateFormatter, YearLocator

indx = pd.date_range('2017-04-01', '2019-01-01')
s = pd.Series(np.random.randn(len(indx)), index=indx)

df = pd.DataFrame(s)

fig, ax = plt.subplots(1)
ax.plot(df)

months = MonthLocator(range(1, 13), bymonthday=1, interval=1)
monthsFmt = DateFormatter("%b")
years = YearLocator(1, month=4, day=1)
yrsFmt = DateFormatter("\n %Y")

ax.xaxis.set_major_locator(years)
ax.xaxis.set_major_formatter(yrsFmt)


ax.xaxis.set_minor_locator(months)
ax.xaxis.set_minor_formatter(monthsFmt)

fig.show()

Also note that I changed %y to %Y so it is formatted as 2017/2018 rather than 17/18.

Answer 2

Matplotlib counts years from zero but UNIX since 1970. Therefore you got years of 48, 49 and etc. To avoid this behavior of matplotlib you have to get from your pandas datetime index date part and then use %Y descriptor to get full years for major ticks:

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.dates import MonthLocator, WeekdayLocator, DateFormatter, YearLocator

indx = pd.date_range('2017-04-01', '2019-01-01')
s = pd.Series(np.random.randn(len(indx)), index=indx.date) # get dates
df = pd.DataFrame(s)

months = MonthLocator() # MonthLocator without args set ticks for every month
monthsFmt = DateFormatter("%b")
years = YearLocator(month=4, day=1)
yrsFmt = DateFormatter("\n%Y") # correct year descriptor

ax = df.plot()
ax.xaxis.set_minor_locator(months)
ax.xaxis.set_minor_formatter(monthsFmt)
for tick in ax.xaxis.get_minor_ticks():tick.label.set_fontsize(9) 
ax.xaxis.set_major_locator(years)
ax.xaxis.set_major_formatter(yrsFmt)

plt.show()