Documentation on Performance considerations

[Dr-Irv]

Problem description

We could use a section in the docs about performance considerations. @TomAugspurger has a nice notebook with suggestions here: http://nbviewer.jupyter.org/github/TomAugspurger/pandas-head-to-tail/blob/master/06-Performance.ipynb

An item to add relates to using .assign(), where, if you have a big DataFrame, it is more inefficient in time and memory to use .assign() as opposed to just creating each new column without the use of method chaining via the paradigm of

    df['newcol1'] = df.a + df.b
    df['newcol2'] = df.c + df.d

[jreback]

see also #3871 and #8178

[andymaheshw]

Working on it at pycon2017 :)

[jorisvandenbossche]

There was also a presentation on PyCon about optimizing pandas: https://github.com/sversh/pycon2017-optimizing-pandas, which can probably be used as well for some inspiration

[mroeschke]

Looks like we have https://pandas.pydata.org/pandas-docs/stable/user_guide/enhancingperf.html so closing for now