Scatterplot of two Pandas Series, coloured by date and with legend

Question

I am studying financial time series, in the format of pandas series. To compare two series I do a scatterplot, and to visualise the time evolution in the scatterplot I can colour them. This is all fine.

My question relates to the legend showing the colours. I would like the legend to show the date/year that the colour corresponds to, rather than just an the index of the data entry, as now. But I haven't been able to do this, or to find a question like this on stackoverflow.

I know the time series knows the dates, and if you just plot a time series, the x-axis will show the dates.

My code is

from pandas_datareader import data as web
import matplotlib.pyplot as plt
import pandas as pd

#Download data
start = '2010-1-1'
end = '2016-10-1'
AAPL = web.DataReader('AAPL', 'yahoo', start=start, end=end)['Adj Close']
TSLA = web.DataReader('GOOG', 'yahoo', start=start, end=end)['Adj Close']

#Scatterplot
plt.scatter(AAPL, TSLA, alpha=.4, c=range(len(AAPL)))
plt.colorbar()
plt.xlabel("AAPL")
plt.ylabel("TSLA")
plt.grid()
plt.show()

This code produces this plot: Scatterplot with colours and legend

Thanks

Answer 1

While there might be an easier answer out there (anyone?), to me the most straightforward way is to change the colorbar ticks manually.

Try the following before you invoke plt.show():

clb = plt.gci().colorbar # get the colorbar artist
# get the old tick labels (index numbers of the dataframes)
clb_ticks = [int(t.get_text()) for t in clb.ax.yaxis.get_ticklabels()]
# convert the old, index, ticks into year-month-day format
new_ticks = AAPL.index[clb_ticks].strftime("%Y-%m-%d")
clb.ax.yaxis.set_ticklabels(new_ticks)

Note that strftime does not appear to be implemented in pandas versions older than 0.18. In that case you'll have to replace .strftime("%Y-%m-%d") with .apply(lambda x: x.strftime('%Y-%m-%d')) as explained here.