Marking specific dates when visualizing a time series

Question

I have a time series that has a few years' worth of data, for example this:

ts = pd.Series(np.random.randn(1000), index=pd.date_range('1/1/2000', periods=1000))

ts = ts.cumsum()

ts.plot()

I also have two extra arrays: let's call the first

dates = [pd.datetime("2000-12-01"), pd.datetime("2001-01-03")]

And the second

labels = ["My birthday", "My dad's birthday"]

labels[i] contains the label for dates[i]. What I'd like to do is to display them in the time series graph so that they can be recognized. One possible visualization could be to display the date on the x axis, draw a vertical line starting from there and have the label either in a legend (with color coding) or somewhere next to the line.

The end result shouldn't be too different from this:

Answer 1

Switching between pandas and matplotlib APIs can be confusing at first.

The solution: get the current axis and then use standard matplotlib API to annotate. This starts you off:

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

ts = pd.Series(np.random.randn(1000),
               index=pd.date_range('1/1/2000',
               periods=1000))

ts = ts.cumsum()
ts.plot()

label_list = [
    (pd.to_datetime("2001-05-01"), 'My\nbirthday', 'r'),
    (pd.to_datetime("2001-10-16"), "Dad's\nbirthday", 'b')
]

ax = plt.gca()

for date_point, label, clr in label_list:
    plt.axvline(x=date_point, color=clr)
    plt.text(date_point, ax.get_ylim()[1]-4, label,
             horizontalalignment='center',
             verticalalignment='center',
             color=clr,
             bbox=dict(facecolor='white', alpha=0.9))

plt.show()

This produces the image below, and you need to look into modifying titles, and text labels and their bounding boxes to the axis object: