如果包含NaN的列如何转换为int？

Question

I am working on this efw data frame:

2   Year    ISO_Code    Countries   INDEX   Rank    Quartile
3   2016    ALB         Albania     7.54    34      1
4   2016    DZA         Algeria     4.99    NaN     4
5   2016    AGO         Angola      5.17    155     4
6   2016    ARG         Argentina   4.84    160     NaN
7   2016    ARM         Armenia     7.57    NaN     1
8   2016    AUS         Australia   7.98    10      1
9   2016    AUT         Austria     7.58    27      NaN

Let's say that I want to convert all the values in the Rank and Quartile columns into integers if the are not a NaN . I have tried efw[["Year", "Rank", "Quartile"]].astype(int) but keep getting this error: ValueError: cannot convert float NaN to integer

Is there a way to tell Python to "skip" the NaN values and only convert the others into int ?

Answer 1

You can convert them into object

efw[["Year", "Rank", "Quartile"]].astype(object)
Out[58]: 
   Year Rank Quartile
0  2016   34        1
1  2016  NaN        4
2  2016  155        4
3  2016  160      NaN
4  2016  NaN        1
5  2016   10        1
6  2016   27      NaN

#efw[["Year", "Rank", "Quartile"]]=efw[["Year", "Rank", "Quartile"]].astype(object)

Answer 2

I have Just reproduced your requirement, here are few way to do this, However i see you have already an integer values in place and only missing ones places as NaN which you can convert into int or say zero as those are missing.

    >>> efw
   Year ISO_Code  Countries  Index Rank Quartile
0  2016      ALB    Albania   7.54   34        1
1  2016      DZA    Algeria   4.99  NaN        4
2  2016      AGO     Angola   5.17  155        4
3  2016      ARG  Argentina   4.84  160      NaN
4  2016      ARM    Armenia   7.57  NaN        1
5  2016      AUS  Australia   7.98   10        1
6  2016      AUT    Austria   7.58   27      NaN

Method 1

Straight from the pandas documentation even we have better way of doing, Document Link For pandas.DataFrame.fillna

So, As you have multiple columns, but only want to replace the NaN in a subset of them, you can use, I would say better would be to replace NaN to 0 rather skipping as blank..

>>> efw.fillna({'Rank':'0', 'Quartile':'0'}, inplace=False)
   Year ISO_Code  Countries  Index Rank Quartile
0  2016      ALB    Albania   7.54   34        1
1  2016      DZA    Algeria   4.99    0        4
2  2016      AGO     Angola   5.17  155        4
3  2016      ARG  Argentina   4.84  160        0
4  2016      ARM    Armenia   7.57    0        1
5  2016      AUS  Australia   7.98   10        1
6  2016      AUT    Austria   7.58   27        0

If you want to make the changes to be permanent in the DataFrame then you can use:

>>> efw.fillna({'Rank':'0', 'Quartile':'0'}, inplace=True)
>>> print(efw)

Method 2

You could check dtype whether it's numeric or not with dtype.kind using apply for your columns as follows..

 *dtype.kind:* A character code (one of 'biufc') identifying the general kind of data. b boolean i signed integer u unsigned integer f floating-point c complex floating-point 

Hence, you can use apply with lambda function as follows ..

 >>> efw.apply(lambda x: x.fillna(0) if x.dtype.kind in 'biufc' else x.fillna('0'))
   Year ISO_Code  Countries  Index Rank Quartile
0  2016      ALB    Albania   7.54   34        1
1  2016      DZA    Algeria   4.99    0        4
2  2016      AGO     Angola   5.17  155        4
3  2016      ARG  Argentina   4.84  160        0
4  2016      ARM    Armenia   7.57    0        1
5  2016      AUS  Australia   7.98   10        1
6  2016      AUT    Austria   7.58   27        0

Method 3

OR With your dataframe you simply use DataFrame.fillna() method. This will convert all NaN into zero irrespective of Columns you have.

>>> efw.fillna(0)
   Year ISO_Code  Countries  Index Rank Quartile
0  2016      ALB    Albania   7.54   34        1
1  2016      DZA    Algeria   4.99    0        4
2  2016      AGO     Angola   5.17  155        4
3  2016      ARG  Argentina   4.84  160        0
4  2016      ARM    Armenia   7.57    0        1
5  2016      AUS  Australia   7.98   10        1
6  2016      AUT    Austria   7.58   27        0

Note: You can use fillna('') rather fillna(0) if you want to remove NaN and make it blank.