CODEWITHSUNDEEP
pythonpandasscikit-learn
Bishwarup Bhattacharjee
Bishwarup Bhattacharjee

Reputation: 582

pandas DataFrame to list of dict using columns as key

I have the following dataframe:

ID     Col1    Col2
1      "A"       "Z"
1      "A"       "Y"
1      "B"       "Z"
2      "A"       "X"
2      "C"       "P"

I want to convert the above in the list of dict form as below with the counts by the ID columns:

[{"A" : 2, "B" : 1, "Z" : 2, "Y" : 1}, {"A" : 1, "C" : 1, "X" : 1, "P" : 1}]

Is there anyway I can achieve that. The dataframe I am having is quite big.

Upvotes: 1

Views: 606

Answers (1)

Alexander
Alexander

Reputation: 109526

Assume your dataframe is named df. You can get the row numbers for each ID using df.groupby('ID').groups:

group_rows = df.groupby('ID').groups

We'll iterate through each group ID, and then use Counter to count the values in Col1 and Col2. I'll then add these to a dictionary.

from collections import Counter

my_dict = {}
for group_id, rows in group_rows.iteritems():
    c = Counter(df.iloc[rows, 1])  # 1 = index number for `Col1`
    c.update(df.iloc[rows, 2])  # 2 = index number for `Col2`
    my_dict[group_id] = dict(c)

>>> my_dict
{1: {'A': 2, 'B': 1, 'Y': 1, 'Z': 2}, 
 2: {'A': 1, 'C': 1, 'P': 1, 'X': 1}}

I chose to output the results to a dictionary instead of your requested list so that the relationship between the group ID and the counted values is explicit. If this is an issue, I can convert them to a list by assuming the dataframe is first sorted by the ID column.

keys = my_dict.keys()
keys.sort()
my_list = [my_dict[k] for k in keys]
>>> my_list
[{'A': 2, 'B': 1, 'Y': 1, 'Z': 2}, {'A': 1, 'C': 1, 'P': 1, 'X': 1}]

Upvotes: 1

Related Questions