How to customize xaxis ticks for DatetimeIndex of Pandas in matplotlib?

Question

I have a pandas dataframe data like this;

            num
Date        
2020-03-01  311
2020-03-08  432
2020-03-15  420
2020-03-22  304
2020-03-29  1030
2020-04-05  1497
2020-04-12  2259
2020-04-19  1938

Where Date is DatetimeIndex . I tried to plot it as below;

from matplotlib import pyplot as plt
x = data.index
fig = plt.figure()
ax1 = fig.add_subplot(111)
tested_plot = ax1.bar(x, data["num"])
dates_fmt = mdates.DateFormatter('%m/%d')
ax1.xaxis.set_major_formatter(dates_fmt)
plt.show()

I wanted the ticks in xaxis to be shown on where the bar exists (like 3/1, 3/8,..., 4/12, 4/19). But when I plot the above, the ticks are reset when April begins and the bars and the ticks are not corresponding;

I know I can convert DatetimeIndex to Index of string using strftime which will solve the problem, but I would like to keep DatetimeIndex for further customization. I also used

import matplotlib.dates as mdates
ax1.xaxis.set_major_locator(mdates.DayLocator(interval=7))

to force ticks to be a week apart, but then I cannot set the start date for ticks and now no bars are on the ticks. Does anybody know how to fix this?

Answer 1

You can try pandas.DataFrame.plot and matplotlib.axes.Axes.set_xticklabels . In the set_xticklabels method, you can format dates as you want (here using pandas.Series.dt.strftime ):

df.plot(kind="bar", ax=ax1)
ax1.set_xticklabels([date.strftime("%Y-%m-%d") for date in df.index], 
                     rotation=45)

---

Full code

# Be sure dates are datetime objects
# df["Date"] = pd.to_datetime(df["Date"])
# df.set_index("Date", inplace=True)

print(df)
#              num
# Date
# 2020-03-01   311
# 2020-03-08   432
# 2020-03-15   420
# 2020-03-22   304
# 2020-03-29  1030
# 2020-04-05  1497
# 2020-04-12  2259
# 2020-04-19  1938

fig = plt.figure()
ax1 = fig.add_subplot(111)
df.plot(kind="bar", ax=ax1)
ax1.set_xticklabels([date.strftime("%Y-%m-%d") for date in df.index], 
                    rotation=45)
plt.show()

output