Errorbar plotting does not always work when the x axis is a datetime:
z = pd.DataFrame({'timestamp': {0: pd.Timestamp('2018-06-16 04:33:27'),
1: pd.Timestamp('2018-06-16 18:07:40')},
'average': {0: 1.4158309812874796, 1: 1.4293226152856995},
'stdev': {0: 0.5721450460404708, 1: 0.5771658975429514}})
Now z is
timestamp average stdev
0 2018-06-16 04:33:27 1.415831 0.572145
1 2018-06-16 18:07:40 1.429323 0.577166
plt.plot(z.timestamp, z.average) works as expected,
but plt.errorbar(z.timestamp, z.average, yerr=z.stdev) produces
<ErrorbarContainer object of 3 artists>
Error in callback <function install_repl_displayhook.<locals>.post_execute at 0x7fe251344840> (for post_execute):
Truncated Traceback (Use C-c C-x to view full TB):
~/.virtualenvs/algorisk/local/lib64/python3.6/site-packages/matplotlib/dates.py in viewlim_to_dt(self)
1024 'often happens if you pass a non-datetime '
1025 'value to an axis that has datetime units'
-> 1026 .format(vmin))
1027 return num2date(vmin, self.tz), num2date(vmax, self.tz)
1028
ValueError: view limit minimum -7.64586229992263e+16 is less than 1 and is an invalid Matplotlib date value. This often happens if you pass a non-datetime value to an axis that has datetime units
What am I doing wrong?
PS. Other values seem to work. E.g.,
plt.errorbar(np.array([datetime.datetime(2018,7,30,12),
datetime.datetime(2018,7,30,15)]).astype("datetime64[h]"),
np.array([2,3]),
yerr=np.array([1,2]))
works as expected.
Natively, matplotlib can plot numpy arrays. Instead of providing pandas Series to the plotting function you probably want to provide the underlying numpy arrays, obtained via .values.
import pandas as pd
import matplotlib.pyplot as plt
df = pd.DataFrame({'timestamp': {0: pd.Timestamp('2018-06-16 04:33:27'),
1: pd.Timestamp('2018-06-16 18:07:40')},
'average': {0: 1.4158309812874796, 1: 1.4293226152856995},
'stdev': {0: 0.5721450460404708, 1: 0.5771658975429514}})
plt.errorbar(df.timestamp.values, df.average.values, yerr=df.stdev.values)
plt.show()