I have a DataFrame of 100M rows and 8 columns. I'm trying to groupby my DataFrame by 5 string columns and perform the following calculations as fast as possible.
df.groupby(['A','B','C','D','E'])['F'].transform('median')
df.groupby(['A','B','C','D','E']).agg({'F':'count', 'G':['mean','median','std'], 'H':['mean','std']})
I'm assuming doing it using numpy arrays would be the fastest, but I don't even know where to begin because it takes a couple of minutes just to convert a column to a numpy array.