使用Pandas Dataframe过滤器功能后的SettingWithCopyWarning

Question

The objective of my code is to overwrite a dataframe with a filtered version. The following code returns the warning beneath:

code:

df = df[df.col>1]

df.col2 = df.col2.astype(float)

error:

/root/.virtualenvs/data_tools/local/lib/python2.7/site-packages/pandas/core/generic.py:2177: SettingWithCopyWarning:
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

theory:

The error occurs on the second line but only happens if the first line is run previously. I believe that the first line is creating a copy of df with the same name as the original which then causes the error. I cannot work out why though.

module versions:

numpy==1.10.1

pandas==0.16.2

Answer 1

The issue as Jeff pointed out is that I was making a view not a copy of the dataframe.

This is what I should have written:

df = df[df.col>1].copy(deep=True)

df.col2 = df.col2.astype(float)