Reshaping pandas DataFrame by transforming columns to labels contained in single column

Question

I am trying to print a df to csv to later export to a Tableau visualization. I can do what I want in Pandas using the labels of my columns, however, to easily work in Tableau, I need to present my data in certain format.

This is my pandas df:

    company   size   year  Australia  Canada  Caribbean  
1      USAA  315.0   1997      200.0   115.0          0          
2     PIMCO  165.0   1997      165.0     0.0          0     
3  Swiss Re   18.0   1997        5.0    10.0        3.0 

I would like to re-shape it so it looks like this more or less:

    company   size   year   value      value 
1      USAA  315.0   1997   Australia  200.0        
2      USAA  315.0   1997   Canada     115.0   
3     PIMCO  165.0   1997   Australia  165.0           
4  Swiss Re   18.0   1997   Australia    5.0
5  Swiss Re   18.0   1997   Canada      10.0       
6  Swiss Re   18.0   1997   Caribbean    3.0

Is there a way to do this easily in Pandas that does not involve creating multiple groupbys?

Answer 1

I had a similar problem, and after trying multiple approaches, this is the most pythonic way I found:

df = pd.melt(df, id_vars=['company', 'size', 'year'])
df = df[df.value != 0]

Hope it helps.