plot major ticks and labels only once per day in Pandas groupby DataFrame

Question

In Pandas, I have a DataFrame of observations (baby bottle feeding volumes) that are indexed by a datetime and grouped by date:

...
bottles = bottles.set_index('datetime')
bottles = bottles.groupby(bottles.index.date)

I want to use matplotlib to plot the cumulative values as they increase each day--that is, show the volume of feedings as it increases each day and resets at midnight:

ax = plt.gca()
ax.xaxis.set_major_locator(mdates.DayLocator())
ax.xaxis.set_minor_locator(mdates.HourLocator())
ax.xaxis.set_major_formatter(mdates.DateFormatter('%d-%m-%Y'))
bottles['volume'].cumsum().plot(kind='bar', figsize=[16,8])
ax.xaxis.grid(True, which="major")
ax.xaxis.grid(False, which="minor")
ax.yaxis.grid(True)
plt.gcf().autofmt_xdate()
plt.show()

Which produces:

I'd like to only label dates on the x-axis once per day, and I'd also like to only draw a vertical grid line on date boundaries (every 24 hours). Any recommendations for how to fix the above code?

pandas assumes that bar plots are categorical plots, so you're always going to have a tick for every bar shown. my guess is thgat you're gonna have to write plot against the matplotlib object directly and not through pandas' interface. — Paul H, May 31 '16 at 15:45

score 0 · Answer 1 · answered Oct 18 '17 at 15:19

Since you didn't provide any data, I generated some dummy data. In essence, you can make the labels invisible by retrieving the ticks on the x-axis, and then making the hourly ticklabels visible.

Note: this works for hours, so resample your dataframe to hours if necessary.

import random
import pandas
import matplotlib.pyplot as plt

#generate dummy data and df
dates = pd.date_range('2017-01-01', '2017-01-10', freq='H')
df = pd.DataFrame(np.random.randint(0, 10, size=(1, len(dates)))[0], index=dates)
ax = df.groupby(pd.TimeGrouper('D')).cumsum().plot(kind='bar', width=1, align='edge', figsize=[16,8]) #cumsum with daily reset.
ax.xaxis.grid(True, which="major")
#ax.set_axisbelow(True)

#set x-labels to certain date format
ticklabels = [i.strftime('%D') for i in df.index]
ax.set_xticklabels(ticklabels)

#only show labels once per day (at the start of the day)
xticks = ax.xaxis.get_major_ticks()
n=24 # every 24 hours
for index, label in enumerate(ax.get_xaxis().get_ticklabels()):
    if index % n != 0:
        label.set_visible(False)  # hide labels
        xticks[index].set_visible(False)  # hide ticks where labels are hidden

ax.legend_.remove()
plt.show()

Result:

plot major ticks and labels only once per day in Pandas groupby DataFrame

1 Answers1