格式化Pandas Dataframe的JSON

Question

I'm trying to wrangle some data to make a recommender system for an app. Of course, to do this I need a record of which users like which posts. I currently have that data in a JSON file that is formatted like this (numbers being post id, and letters being user ids):

    {
       "-1234": {
         "abc": "abc",
         "def": "def",
         "ghi": "ghi"
    },
       "-5678": {
         "jkl": "jkl",
         "mno": "mno"
    }

I'm trying to figure out how to get this into a pandas dataframe that would look like this:

example format

I've tried using a few online JSON to CSV converters out of laziness which unsurprisingly didn't bring it into a useable format for me. I've tried using "print(json_normalize(data))", as well which also did not work, and put each instance of a like into separate columns.

Any advice?

Answer 1

From my experience for such simple formats, writing a quick and dirty loop is usually the fastest method rather than finding some ready solution and customizing it. An example for the data you gave here:

import json
my_json="""    {
       "-1234": {
         "abc": "abc",
         "def": "def",
         "ghi": "ghi"
    },
       "-5678": {
         "jkl": "jkl",
         "mno": "mno"
    }
    }"""
parsed_json = json.loads(my_json)
print(parsed_json)
# result:
# {'-1234': {'abc': 'abc', 'def': 'def', 'ghi': 'ghi'},
# '-5678': {'jkl': 'jkl', 'mno': 'mno'}}

for key in parsed_json.keys():
    line = ''
    line += key
    line += ' | '
    for value in parsed_json[key].values():
        line += value + ', '
    line = line[:-2] # stripping the ', ' from the end of the line
    print(line)
# result:
# -1234 | abc, def, ghi
# -5678 | jkl, mno

Answer 2

This is a solution optimized for the peculiarities in your dataset.

import pandas as pd
data = {
       "-1234": {
         "abc": "abc",
         "def": "def",
         "ghi": "ghi"
    },
       "-5678": {
         "jkl": "jkl",
         "mno": "mno"
    }}
formatted = [{'PostID': d, 'User Like': list(data[d].keys())} for d in data]
df = pd.DataFrame.from_dict(formatted)

Output:

Answer 3

Setup

Thanks Zaroth

import json
my_json="""    {
       "-1234": {
         "abc": "abc",
         "def": "def",
         "ghi": "ghi"
    },
       "-5678": {
         "jkl": "jkl",
         "mno": "mno"
    }
    }"""
parsed_json = json.loads(my_json)

---

Comprehension

pd.DataFrame(
    [(k, [*v]) for k, v in parsed_json.items()],
    columns=['PostID', 'User Like']
)

  PostID        User Like
0  -1234  [abc, def, ghi]
1  -5678       [jkl, mno]

OR

pd.DataFrame({
    'PostID': [*parsed_json],
    'User Like': [[*v] for v in parsed_json.values()]
})

Answer 4

data = {"-1234": {"abc": "abc","def": "def","ghi": "ghi"},"-5678": {"jkl": "jkl","mno": "mno"}}

key = []
val = []

for k,v in data.items():
    key.append(k)
    val.append(list(v.values()))

pd.DataFrame(zip(key,val),columns=['PostID','User Like'])