CODEWITHSUNDEEP
pythonpandaspandas-groupby
Niccola Tartaglia
Niccola Tartaglia

Reputation: 1667

Categorical Variables pandas

I would like to compare stats such as mean,std etc of my dataset conditional on a dummy variable. I saw a command for that somewhere but I cannot remember (or google search was not succesful). I would like to produce an output like this:

dummy    mean(var1)  mean(var2)
0          1.5        3
1          10         10

Maybe something with groupby??

This is a min example:

    dict1 = [{'dummy': '0', 'var1': 1, 'var2': 2},
         {'dummy': '0', 'var1': 2, 'var2': 4},
        {'dummy': '1', 'var1': 5, 'var2': 8},
        {'dummy': '1', 'var1': 15, 'var2': 12},]
 df = pd.DataFrame(dict1, index=['s1', 's2','s3','s4'])

Upvotes: 1

Views: 73

Answers (1)

rafaelc
rafaelc

Reputation: 59274

I believe you want groupby+describe

ndf = df.groupby('dummy').describe()

Then just select whatever info you want

ndf.loc[:, ndf.columns.get_level_values(1)=='mean']

       var1     var2
       mean     mean
dummy       
0      1.5      3.0
1      10.0     10.0

describe might be more powerful because will give you lots of different stats upfront. But of course you can select your methods, .e.g

df.groupby('dummy').mean()


       var1    var2
dummy       
0      1.5      3.0
1      10.0    10.0


df.groupby('dummy').std()


       var1      var2
dummy       
0      0.707107  1.414214
1      7.071068  2.828427

Upvotes: 2

Related Questions