DataFrame replace slow on DataFrames containing strings when using use_inf_as_null option

[spillz]

Probably an edge case but something that bit me

import pandas, numpy as np
df = pandas.DataFrame({'a':[0]*5000+[1]*5000, 'b':[2]*5000+[1]*5000 , 'c': ['a']*5000 + ['b']*5000})
%timeit df.replace(1,3, inplace=True)

df = pandas.DataFrame({'a':[0]*5000+[1]*5000, 'b':[2]*5000+[1]*5000 , 'c': ['a']*5000 + ['b']*5000})
def rep(df):
    for c in df.columns:
        df.loc[df[c]==1,c] = 3
    return df
%timeit rep(df)

pandas.set_option('use_inf_as_null', True)
df = pandas.DataFrame({'a':[0]*5000+[1]*5000, 'b':[2]*5000+[1]*5000, 'c': ['a']*5000 + ['b']*5000})
%timeit df.replace(1,3, inplace=True)

df = pandas.DataFrame({'a':[0]*5000+[1]*5000, 'b':[2]*5000+[1]*5000 , 'c': ['a']*5000 + ['b']*5000})
def rep(df):
    for c in df.columns:
        df.loc[df[c]==1,c] = 3
    return df
%timeit rep(df)

One of these things is not like the other!

1000 loops, best of 3: 1.77 ms per loop
100 loops, best of 3: 5.89 ms per loop
1 loop, best of 3: 2.24 s per loop
100 loops, best of 3: 5.92 ms per loop

Pandas 0.20.1

[jreback]

in 0.21.0

this looks ok.

494 us +- 5.83 us per loop (mean +- std. dev. of 7 runs, 1000 loops each)
8.1 ms +- 553 us per loop (mean +- std. dev. of 7 runs, 100 loops each)
523 us +- 33.2 us per loop (mean +- std. dev. of 7 runs, 1000 loops each)
7.54 ms +- 212 us per loop (mean +- std. dev. of 7 runs, 100 loops each)

can you give a try. also would take a PR with asv's for this.

[jreback]

note that the option changed in 0.21.0 (same just null -> na)

pandas.set_option('use_inf_as_na', True)

[mroeschke]

Given that this option was deprecated in 2.1 #53494, going to close as a wont fix