如何将数据框拆分为两行，其中两列具有特定值？

Question

I have a dataframe which contains information by column such as:

Month  Year     Cost_1 Cost_2
1      2017     100    0
2      2017     0      100
3      2017     140    30

and I am looking to transpose this data so that it takes the form:

Month  Year     Cost_1 Cost_2 Type
1      2017     100    0      Cost_1
2      2017     0      100    Cost_2
3      2017     140    0      Cost_1
3      2017     0      30     Cost_2

My initial thought was to use .loc(Cost_1>0,"Type")="Cost_1" but this wouldn't deal with the rows which have both Cost_1 and Cost_2 and need a new row adding? Should I split the data so that it has only Cost_1 or Cost_2 first and then use .loc to create the Type column or is there a smarter way to do this?

Edit:

The problem is actually more complicated than I first thought. Each column has an associated partner Cost_1 has Count_1, Cost_2 has Count_2.. etc.

   Year  Month BDADExclIncurred_Capped_count  BDADExclIncurred_Capped_mean  BDTPDIncurred_Capped_count BDTPDIncurred_Capped_mean
0  2015      5                             0                         NaN                          60                         900
1  2015     10                             0                         NaN                           0                         NaN 
2  2015     12                             0                         NaN                           0                         NaN 
3  2016      1                            60                        2000                           0                         NaN 
4  2016      1                           100                        1500                          20                         600 

This is how my data looks before, with many columns broken up into count:mean pairs, I want to keep those together but if there is a row with two count:mean pairs I want that to be split into two rows, where each has only one corresponding count:mean pair. Then I wish to create a new column called "type" which tells me what the count:mean pair associated with that row is.

   Year  Month BDADExclIncurred_Capped_count  BDADExclIncurred_Capped_mean  BDTPDIncurred_Capped_count BDTPDIncurred_Capped_mean Type
0  2015      5                             0                         NaN                          60                         900  TPD
1  2015     10                             0                         NaN                           0                         NaN  
2  2015     12                             0                         NaN                           0                         NaN  
3  2016      1                            60                        2000                           0                         NaN  AD
4  2016      1                           100                        1500                           0                           0  AD
5  2016      1                             0                           0                          20                         600  TPD

As show in this example, a new row is created. Index 4 from the previous dataframe is now split into both index 4 and index 5.

Answer 1

Assuming either only Cost_1 or Cost_2 are greater than zero, as your example suggests, here's an simple approach to populate Type with Cost_1 and Cost_2 in one step:

c = ['Cost_1','Cost_2']
counts = df[c].gt(0).dot(df[c].columns + ',').str.rstrip(',').str.split(',')
counts_df = pd.DataFrame(counts.tolist(), columns = ['Count_1', 'Count_2'])
df.assign(**counts_df)

    Month  Year  Cost_1 Count_1  Cost_2 Count_2
0      1  2017     100  Cost_1       0       0
1      2  2017       0  Cost_2     100       0
2      3  2017     140  Cost_1      30  Cost_2