Groupby with some condition (Pandas)

Question

for instance, I have this sample dataframe

    Depth  Fluid

0   235.5  nan
1   236    water
2   236.5  water
3   237    nan
4   237.5  water
5   238    water

Now i want to get the sample data just to be like this

    Min_Depth  Max_Depth  Fluid

0   236        236.5      water
1   237.5      238        water

How do I make it?

Asked By: nrasif18

||

Source

Answer 1

You can check with cumsum

x = df.Fluid.ne('water')
out = df[~x].groupby([x.cumsum(),df.Fluid]).agg(max_dp= ('Depth','max'),
                                                min_dp= ('Depth','min')).reset_index(level=1)
out
Out[202]: 
       Fluid  max_dp  min_dp
Fluid                       
1      water   236.5   236.0
2      water   238.0   237.5

Answered By: BENY

Answer 2

Another possible solution, based on pandas.unstack:

(df.dropna()
 .assign(Fluid = lambda x: sorted(list(range(len(x)//2)) * 2))
 .set_index(['Fluid', ['minD', 'maxD'] * 2])
 .unstack()
 .droplevel(0, axis=1)
 .reset_index()
 .assign(Fluid = 'water'))

Output:

   Fluid   maxD   minD
0  water  236.5  236.0
1  water  238.0  237.5

Answered By: PaulS

Groupby with some condition (Pandas)

Question:

Answers: