Converting Groupby Object into DataFrame objects in a loop

Question

I have a dataframe df that looks like this

No Country Sex Age Group
1    Spain   M Young
2    Norway  F Middle
3    Mexico  M Elderly
...

My aim is to group this data based on country first, then do analysis based on sex and age groups, using .value_counts()

Thus, I used groupby on this via df2 = df.groupby(df.Country)and looped like this

for d in df2:
    print(type(d))

I am getting this output:

<class 'tuple'>
<class 'tuple'>
<class 'tuple'>
...

I wanted to use

for d in df2:
    print(pd.DataFrame(d).Sex.value_counts())

but I am getting this error

DataFrame constructor not properly called!

I had asked a similar question yesterday, but that was limited to a very particular query. What I want to learn is to know how to "group" layers into chunks and then do analysis on those particular chunks.

Thanks!

the reason you get <class 'tuple'> is because .groupby() returns a tuple of size 2, where the first element is your grouping (i.e. the country) and the second is the data for that group, which is already a dataframe. — asongtoruin
– asongtoruin, Commented Jun 28, 2018 at 8:08

jezrael · Accepted Answer · 2018-06-28 07:50:29Z

2

I think better is use:

s = df.groupby('Country').Sex.value_counts()
print (s)
Country  Sex
Mexico   M      1
Norway   F      1
Spain    M      1
Name: Sex, dtype: int64

But if want use loops is necessary unpack tuples name of group and df by adding another variable i:

df2 = df.groupby(df.Country)
for i, d in df2:
    print(type(d))
<class 'pandas.core.frame.DataFrame'>
<class 'pandas.core.frame.DataFrame'>
<class 'pandas.core.frame.DataFrame'>

Or use GroupBy.apply:

def func(x):
    print (x)
    a = x.Sex.value_counts()
    #another code 
    return a

df2 = df.groupby(df.Country).apply(func)

answered Jun 28, 2018 at 7:50

jezrael

868k103 gold badges1.4k silver badges1.3k bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Vibhu Over a year ago

Thanks! Groupby.apply is most useful to me!

jezrael Over a year ago

@Vibhu - Yes, it is obviously nice use, but there is one gotcha, see Warning

Collectives™ on Stack Overflow

Converting Groupby Object into DataFrame objects in a loop

1 Answer 1

2 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related