熊猫中几列的汇总

Question

I have the following data in dataframe df:

VALUE   COUNT   REGION  ID
235     15      LP      139
355     59      LP      102
421     8       LP      127
427     227     LP      90
439     4       LP      133
235     45      UP      139
355     231     UP      102
421     756     UP      127
427     23      UP      90
439     76      UP      133

I want to combine the data frame such that the rows with the common 'VALUE' are merged. Their count should be added up. However, the column REGION need not be included in the final dataframe. I tried the following:

df.groupby(['VALUE'])['COUNT'].sum()

How do I get it to return all columns (except REGION)?

Answer 1

You can tell aggregate to perform multiple actions on multiple columns.

You did not mention what you want to do with the ID column, so here I take the first. Columns that can't be summed are usually silently dropped, and so is the case here.

In [51]: df.groupby('VALUE').aggregate({'COUNT':np.sum, 'ID':lambda x:x.iloc[0]})
Out[51]: 
       COUNT   ID
VALUE            
235       60  139
355      290  102
421      764  127
427      250   90
439       80  133

(in general, the docs of groupby is one of the most useful pieces of info you'll read regarding pandas imho)