How to make this work? df['sum_greater_then'] = df.groupby(['scan_number', 'raw_file]).sort_values('rank', ascending=False)['intensity'].cumsum()

I need to normalize the data. For this I first need to add intensities up based on their rank in the spectra. One spectra is when 'scan_number' and 'raw_file' don't change.

These are the calculations, which I need to do:

Rank 1: (4,03 2,05 1,057) / total = 7,137/7,137 = 1

Rank 2: (2,05 1,057) / total = 0,435

Rank 3: 1,057 / total = 0,148

There are other columns in the dataframe, I don't need them for this specific calculation though.

I already ranked each intensity, calculated the total and created a column each for those values. I just need to add up the right intensities to divide them by total and get the normalized value in the end.

This doesn't work as well df['sum_greater_then'] = df.groupby(['scan_number', 'raw_file]).apply(lambda x: x.sort_values('rank', ascending=False)['intensity'].cumsum())

CodePudding user response：

This is a little clunky, but I believe achieves what you want.