How do I change the density of x-ticks of a pandas time series plot?

Question

I am trying to generate a smaller figure visualising a pandas time series. The automatically-generated x-ticks, however, do not adapt to the new size and result in overlapping ticks. I am wondering how can I adapt the frequency of the x-ticks? Eg for this example:

figsize(4, 2)

num = 3000
X = linspace(0, 100, num=num)
dense_ts = pd.DataFrame(sin(X) + 0.1 * np.random.randn(num),
                        pd.date_range('2014-01-1', periods=num, freq='min'))

dense_ts.plot()

The figure that I get is:

I am able to work around this problem using the Matplotlib date plotting, but it is not a very elegant solution - the code requires me to specify all the output formatting on a per-case basis.

figsize(4, 2)

from matplotlib import dates

fig, ax = plt.subplots()
ax.plot_date(dense_ts.index.to_pydatetime(), dense_ts, 'b-')
ax.xaxis.set_minor_locator(dates.HourLocator(byhour=range(24),
                                           interval=12))
ax.xaxis.set_minor_formatter(dates.DateFormatter('%H:%m'))
ax.xaxis.set_major_locator(dates.WeekdayLocator())
ax.xaxis.set_major_formatter(dates.DateFormatter('\n\n%a\n%Y'))

plt.show()

I'm wondering if there is a way to solve this issue using the pandas plotting module or maybe by setting some axes object properties? I tried playing with the ax.freq object, but couldn't really achieve anything.

Answer 1

You can pass a list of x axis values you want displayed in your dense_ts.plot()

dense_ts.plot(xticks=['10:01','22:01'...])

Another example for clarity

df = pd.DataFrame(np.random.randn(10,3))

Plot without specifying xticks

df.plot(legend=False)

Plot with xticks argument

df.plot(xticks=[2,4,6,8],legend=False)