Covert groups to json after groupby() in Pandas-CodePudding

I want to convert groups to dict after groupby function in pandas. This is the data sample:

  user  val1   val2
0    A     1    one
1    A     3    one
2    A     2    one
3    B     2    two
4    B     2  three
5    C     3    one

After

df = df.groupby('user')

I want to get result something like this:

group_A = {'val1':[1,3,2], 'val2':['one', 'one', 'one']}
group_B = {'val1':[2,2], 'val2':['two', 'three']}
group_C = {'val1':[3], 'val2':['one']}

So for example:

for group in df:
    print(group)

Result:

{'val1':[1,3,2], 'val2':['one', 'one', 'one']}
{'val1':[2,2], 'val2':['two', 'three']}
{'val1':[3], 'val2':['one']}

CodePudding user response：

So you can do

df.groupby('user').agg(lambda x : x.values.tolist()).to_dict('records')
Out[165]: 
[{'val1': [1, 3, 2], 'val2': ['one', 'one', 'one']},
 {'val1': [2, 2], 'val2': ['two', 'three']},
 {'val1': [3], 'val2': ['one']}]

d = df.groupby('user').agg(lambda x : x.values.tolist()).to_dict('index')
d['A']

If do need for loop

for x, y in df.groupby('user'):
    print(y.drop(['user'],1).to_dict('list'))

CodePudding user response：

So, i came up with the same idea of Beny but with a function to store it in a variable :

group_A, group_B, group_C = {},{},{}

def add_to_dict (user_dict,user): 
  d = df.groupby('user').agg(lambda x : x.values.tolist()).to_dict('index')
  return (d[user])

group_A = add_to_dict(group_A,'A')
group_B = add_to_dict(group_B,'B')
group_C = add_to_dict(group_C,'C')