Parsing JSON to do math?

I'd recommend using a dedicated package for calculations like pandas:

first_exam_grades = pd.DataFrame.from_dict(first_exam_results, orient='index').astype(int)second_exam_grades = pd.DataFrame.from_dict(second_exam_results, orient='index').astype(int)improvements = second_exam_grades.LastExamGrade.to_frame()improvements['Improvement'] = second_exam_grades.LastExamGrade - first_exam_grades.LastExamGrade

This will give you something that looks like this:

Now you can output it anyway you'd like

list(zip(*([improvements.index.tolist()] + [improvements[c].values.tolist() for c in improvements])))

This will give you [('Bob', 54, 9), ('Jack', 70, 12)] as you want.

One possible solution, using coroutines. Coroutine receive_message holds up to last two values LastExamGrade from the message for each student and produces list of student name, last grade and improvement over last grade:

json_messages = [# 1st message:{   "Bob":  {  "Age": "15",  "LastExamGrade": "45",  },    "Jack":  {  "Age": "16",  "LastExamGrade": "58",  }},# 2nd message{    "Bob":  {  "Age": "15",  "LastExamGrade": "54",  },    "Jack":  {  "Age": "16",  "LastExamGrade": "70",  }},# 3nd message (optional){    "Bob":  {  "Age": "15",  "LastExamGrade": "14",  },    "Jack":  {  "Age": "16",  "LastExamGrade": "20",  }}]def receive_message():    d, message = {}, (yield)    while True:        for k, v in message.items():            d.setdefault(k, []).append(v['LastExamGrade'])            d[k] = d[k][-2:] # store max last two messages        message = yield [(k, *tuple(v if len(v)==1 else [v[1], str(int(v[1])-int(v[0]))])) for k, v in d.items()]receiver = receive_message()next(receiver) # prime coroutinefor msg in json_messages:    print(receiver.send(msg))

Prints:

[('Bob', '45'), ('Jack', '58')][('Bob', '54', '9'), ('Jack', '70', '12')][('Bob', '14', '-40'), ('Jack', '20', '-50')]