python astype(str) gives SettingWithCopyWarning and requests I use loc

Question

Using this simple line of code, I keep on getting a SettingWithCopyWarning error that than carries through my whole code.

#make email a string
df['Email Address'] = df['Email Address'].astype(str)

C:\Users\xxx\AppData\Local\Continuum\Anaconda2\lib\site-packages\ipykernel\__main__.py:2: 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

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy
  from ipykernel import kernelapp as app

I went through the documentation, but can't make it work with loc. The code below is wrong.

df.loc['Email Address'] = df.loc['Email Address'].astype(str)

Please excuse if this is a duplicate question - I searched it on stackoverflow, but couldn't find one that addresses loc and astype.

Do, ``df.ix[:, 'Email Address'] = df['Email Address'].astype(str)`` — Meitham, Jul 11 '17 at 14:59

score 22 · Accepted Answer · answered Jul 11 '17 at 14:52

22

Your issue isn't with how you are making the assignment. It is with the dataframe prior to assignment. At some point prior to the assignment, you created df in such a way that it became a view into another dataframe. You can verify this with bool(df.is_copy)

If you are ok with df being a separate thing with no linkages to data in other dataframes...

df = df.copy()

Then proceed to make your assignment.

answered Jul 11 '17 at 14:52

piRSquared

285,575
57
475
624

Wow - that's correct, I did create a new dataframe from a filter of a previous dataframe df = raw[raw['COUNTRY'].isin(['United States', 'Canada']) - is the issue with how I did that or do I need to do the df = df.copy() no matter what? Thank you – jeangelj Jul 11 '17 at 15:07
Why should this trigger the warning, anyway? Would doing `df = df.copy()` increase RAM usage since it is no longer just a "view"? When you do `.astype()` when `bool(df.is_copy)==True`, is it secretly performing `.copy()` under the hood and therefore the warning is triggered, since it copies the df without your explicit telling it to do so? – Corey Levinson May 13 '19 at 20:30
Yes, there are many ways to trigger a copy. The point is that it is inconsistent and difficult to predict. `df.copy()` is a pass to just get the copy without messing around. – piRSquared May 13 '19 at 20:33

score 2 · Answer 2 · answered Mar 16 '21 at 19:37

2

Update 03/21

I believe this is the correct solution with loc

df.loc[:, 'Email Address'].astype(str)

answered Mar 16 '21 at 19:37

Pysnek313

134
14

python astype(str) gives SettingWithCopyWarning and requests I use loc

2 Answers2

Linked