df.ix method changed results between 0.10.1 to 0.11.0

[jobingr]

I have a big DataFrame where i want to to overwrite values in a given column if the values in another column is not in a list. Value to be overwritten is from the checked column. The following code gives the expected results in 0.10.1, but gives unexpected results in 0.11.0

using numpy 1.7.1

code:

df contains the original pandas DataFrame

accurate_list = ['SA EQUITY CFD', 'SA EQUITY', 'SA SSF']

x is the df of items not matching the list in its Classification

x = df[~df.Classification.isin(accurate_list)]

overwrites only incorrect Classifications

df.ix[x.index,'Industry Code'] = df['Classification']

I have been unable to recreate the error with smaller dataframes. Have double checked and the dataframe definitely has an unique index. No error message, just incorrect values written to the 'Industry Code' column

[jobingr]

sorry... code got markedown applied

[jreback]

we need to see a sample that reproduces this problem...

and try on master, have been some bug fixes in this area

[jobingr]

have not checked on 0.10.1 but apparently fixed there.

http://stackoverflow.com/questions/17020763/pandas-0-10-1-to-0-11-0-ix-method

[jreback]

your example is indicted in #3839 and is fixed in #3670
thanks for the report