Pandas 0.15.1 df.interpolate().plot() wrong result with NaT in datelike index

[sorenwacker]

I am using a DataFrame df with dates as index and some data with missing values.
All dates are from the same year 1999 and the data has some missing vales.
There is one record in the dates that is NaT.

When I use df.interploate().plot() a line is plotted between 1996 and 1716.
When NaT is committed the correct plot appears.

[jreback]

you need to provide copy pastable code and pd.show_versions()

[sorenwacker]

import pandas as pd
import numpy as np
pd.options.display.mpl_style = 'default'
import matplotlib.pyplot as plt
%pylab inline
data=np.random.rand(5)
data[4]=NaN
times=['19990103','19990106','19990109','19990112','NaT']
dftimes = pd.to_datetime(times,format='%Y%m%d')
for i in dftimes:
print i
df=pd.DataFrame(data,dftimes)
df.interpolate().plot()

INSTALLED VERSIONS

commit: None
python: 2.7.5.final.0
python-bits: 64
OS: Linux
OS-release: 3.17.2-200.fc20.x86_64
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8

pandas: 0.15.1
nose: 1.3.4
Cython: 0.21
numpy: 1.9.1
scipy: 0.14.0
statsmodels: 0.5.0
IPython: 2.2.0
sphinx: 1.2.3
patsy: 0.3.0
dateutil: 1.5
pytz: 2014.9
bottleneck: None
tables: 3.1.1
numexpr: 2.3.1
matplotlib: 1.4.2
openpyxl: 1.8.5
xlrd: 0.9.3
xlwt: 0.7.5
xlsxwriter: 0.5.7
lxml: 3.4.0
bs4: 4.3.2
html5lib: None
httplib2: 0.7.7
apiclient: None
rpy2: 2.5.2
sqlalchemy: 0.9.7
pymysql: None
psycopg2: None

[jreback]

This is correct. Not sure what you think is incorrect here. Not sure that matplotlib handles NaT in plotting. You will generally need to drop these indexes. As I pointed out before it is a little odd to do this (have NaT in the index).

In [29]: df.interpolate()
Out[29]: 
                   0
1999-01-03  0.791905
1999-01-06  0.406823
1999-01-09  0.300865
1999-01-12  0.690322
NaT         0.690322

In [32]: df.interpolate(method='index')
NotImplementedError: Interpolation with NaNs in the index has not been implemented. Try filling those NaNs before interpolating.

[sorenwacker]

look at the plot

df.interpolate().plot()

there you see that the line expanded to the 18th century.

[jreback]

I am using matplotlib 1.4.0 and current master of pandas, and it raises
ValueError: No sensible date limit could be found in the AutoDateLocator.

seems ok to me.

Not really sure matplotlib actually supports NaT on an index (as I indicated above).
And in general having a missing value on an INDEX is just not a good idea. doesn't make much sense.

and easy enough to simply drop them first.

@TomAugspurger

[TomAugspurger]

Agreed that plotting (or anything) with a NaT in the index isn't good practice.

Although strangely, these two aren't the same:

In [52]: df.interpolate().plot()
Out[52]: <matplotlib.axes._subplots.AxesSubplot at 0x10daf7a90>

In [53]: fig, ax = plt.subplots()

In [54]: ax.plot(df.index, df.interpolate().values)

The first one has the x-axis decreasing (most recent to oldest)

The second one has the x-axis increasing.

[jondo]

Could the currently processed #12405 be a duplicate of this?

[jreback]

@jondo could be; i'll put an xref on that issue.