On 2018-05-15 06:23, Zoran Ljubišić wrote:
> Matt,
>
> thanks for the info about pydata mailing group. I didn't know it exists.
> Because comp.lang.python is not appropriate group for this question, I
> will continue our conversation on gmail.
>
> I have put len(df.CRM_assetID.cat
> <http://df.CRM_assetID.cat>.categories.isin(['V1254748', 'V805722',
> 'V1105400'])) = 55418 in next message, after I noticed that this
> information is missing.
>
> If I want to select all rows that have categories from the list, how
> to do that?
>
> Regards,
>
> Zoran
>
Hi Zoran-
(Including python-list again, for lack of a reason not to. This
conversation is still relevant and appropriate for the general Python
mailing list -- I just meant that the pydata list likely has many more
Pandas users/experts, so you're more likely to get a better answer,
faster, from a more specialized group.)
Selecting all rows that have categories is a bit simpler than what you
are doing -- your issue is that you are working with the *set of
distinct categories*, and not the actual vector of categories
corresponding to your data.
You can select items you're interested in with something like the following:
"""
In [1]: import pandas as pd
In [2]: s = pd.Series(['apple', 'banana', 'apple', 'pear', 'banana',
'cherry', 'pear', 'cherry']).astype('category')
In [3]: s
Out[3]:
0 apple
1 banana
2 apple
3 pear
4 banana
5 cherry
6 pear
7 cherry
dtype: category
Categories (4, object): [apple, banana, cherry, pear]
In [4]: s.isin({'apple', 'pear'})
Out[4]:
0 True
1 False
2 True
3 True
4 False
5 False
6 True
7 False
dtype: bool
In [5]: s.loc[s.isin({'apple', 'pear'})]
Out[5]:
0 apple
2 apple
3 pear
6 pear
dtype: category
Categories (4, object): [apple, banana, cherry, pear]
"""
(Note that I'm also passing a set to `isin` instead of a list -- this
doesn't matter when looking for two or three values, but if you're
passing 1000 values to `isin`, or 10_000, or 1_000_000, then linear-time
membership testing can start to become an issue.)
You are accessing the vector of the *unique categories* in that column, like
"""
In [6]: s.cat.categories
Out[6]: Index(['apple', 'banana', 'cherry', 'pear'], dtype='object')
In [7]: s.cat.categories.isin({'apple', 'pear'})
Out[7]: array([ True, False, False, True])
"""
The vector `s.cat.categories` has one element for each distinct category
in your column, and your column apparently contains 55418 different
categories.
MMR...
--
https://mail.python.org/mailman/listinfo/python-list
