I would like to use the modern way of Pandas method chaining to assign values to a subset of a column.
Let's say I have the following dataframe
df = pd.DataFrame({'a': [1, 0, 0, 1]})
a
0 1
1 0
2 0
3 1
I would like to achieve the equivalent of
df.loc[df.a == 1, 'a'] = 2
with something like
df.query('a == 1').assign(a=2)
However, the above creates a subset dataframe and does not modify the whole dataframe. Is that somehow possible to achieve?
The query method, as its name indicates, is designed for querying a dataframe, not for setting values.
As such, loc is entirely appropriate, noting you can assign to a series via a string:
df.loc[df.a == 1, 'a'] = 2
More idiomatic may be to use pd.Series.mask, which you can even use in place:
df['a'].mask(df['a'] == 1, 2, inplace=True)
You should view "method chaining" as a means to an end rather than a requirement or objective in its own right. If you are set on using method chaining, you can use pd.DataFrame.assign:
df = df.assign(a=df['a'].mask(df['a'] == 1, 2))
As an isolated operation, I find this less readable. But you may find it useful with multiple linked operations via method chaining.