In order to decide as to whether some operations return a copy or a view, I consulted the posts to the question: What rules does Pandas use to generate a view vs a copy?
It reads as follows:
Here's the rules, subsequent override:
All operations generate a copy
If inplace=True is provided, it will modify in-place; only some operations support this
An indexer that sets, e.g. .loc/.iloc/.iat/.at will set inplace.
An indexer that gets on a single-dtyped object is almost always a view (depending on the memory layout it may not be that's why this is not reliable). This is mainly for efficiency. (the example from above is for .query; this will always return a copy as its evaluated by numexpr)
An indexer that gets on a multiple-dtyped object is always a copy.
import pandas as pd
import numpy as np
df = pd.DataFrame({'A': 'foo bar foo bar foo bar foo foo'.split(),
'B': 'one one two three two two one three'.split(),
'C': np.arange(8), 'D': np.arange(8) * 2})
df2 = df.loc[0,]
df2[0] = 500
So this last line does not change df. Reasoning: The call "df.loc[]" is an indexer that gets on a multi-dtyped object. Therefore df2 is a copy. Now, I thought I got the rules - however when using the following:
df3 = df.iloc[0:2,]
df3[:] = 55
I get a) a warning saying "A value is trying to be set on a copy of a slice from a DataFrame." and b) the original df is changed - but only partially. The first two columns are unchanged, whereas the last two are changed to 55.
I don't understand this behavior wrt the rules outlined above. For instance, "df.iloc[0:2,]" is an indexer that gets on a multi-dtyped object and should therefore return a copy. So why do I get the warning that a value is set on a copy of a slice?
