Question

Suppose I have two DataFrames a & b where a is larger than b and has all NaNs. b.index is a subset of a.index, however b has real values. I wish to merge the values from b into a.

In [102]: mset

Out[102]: 
<class 'pandas.core.frame.DataFrame'>
DatetimeIndex: 9446 entries, 2012-11-02 07:00:00 to 2012-11-05 15:24:00
Data columns:
open     9207  non-null values
high     9207  non-null values
low      9207  non-null values
close    9207  non-null values
dtypes: float64(4)

In [103]: a
Out[103]: 
<class 'pandas.core.frame.DataFrame'>
DatetimeIndex: 1440 entries, 2012-11-14 00:00:00 to 2012-11-14 23:59:00
Freq: T
Data columns:
open     0  non-null values
high     0  non-null values
low      0  non-null values
close    0  non-null values
dtypes: float64(4)

There’s an example of what the dataframes look like.

EDIT:

I’d also like to retain the index as well

Answer 1

Your example dataframes do not match the statement “b.index is a subset of a.index, however b has real values.” The dates are not overlapping. However:

This works for integer indexes, don’t know about ts indexes:

a.ix[a.index] = mset.ix[a.index]