如何合并字典列表并将其转换为Python中的多个数据框？

Question

I have several lists of dictionaries and I want to convert them into dataframes. I first used update to convert a list of dictionaries into a dictionary of dictionaries and then used pd.concat to concatenate each dictionary.

I grouped data by hospital ID and each list has two dictionaries. Within each dictionary, there's dataframe with columns 'hospital' , 'patientID' , and 'results'

# Hospital35006 Adults Test results
diabetes_35006 =
[{'hospital': [35006, 35006], 'patientID': [0001, 0002], 'results': [0,1]}] #Adult Patients(18-25yrs) 
[{'hospital': [35006, 35006], 'patientID': [0003, 0004], 'results': [1,0]}] #Adult Patients(25-30yrs)


# Hospital35007 Adults Test results
diabetes_35007 = 
[{'hospital': [35007, 35007], 'patientID': [0001, 0002], 'results': [0,1]}] #Adult Patients(18-25yrs)
[{'hospital': [35007, 35007], 'patientID': [0003, 0004], 'results': [1,0]}] #Adult Patients(25-30yrs)


def resultDF(test_results):
    adults_test_results = {} 
    for results in test_results:
        adults_test_results.update(results) #Concatenate two adults test results in diabetes_35006 & diabetes_35007 
    dataframe = pd.concat(adults_test_results, ignore_index = True) 
    return dataframe

hospital_35006 = resultDF(diabetes_35006)
hospital_35007 = resultDF(diabetes_35007)

Since I also have test results from another 10 hospitals, is there something I can add to my code to generate dataframes more efficiently rather than writing hosiptal_35006 = resultDF(diabetes_35006) ... etc. each time?

Answer 1

I think best practice would be to have a dictionary of dictionaries ("diabetes") and then convert it to a dictionary of dataframes ("hospital") using dictionary comprehension.

This link can be useful to avoid the temptation of dynamically generating variables based on string: http://stupidpythonideas.blogspot.co.uk/2013/05/why-you-dont-want-to-dynamically-create.html

I assume (?) the correct input you have is (list of dicts):

# Hospital35006 Adults Test results
diabetes_35006 =\
[{'hospital': [35006, 35006], 'patientID': [0001, 0002], 'results': [0,1]},\
#Adult Patients(18-25yrs)\
{'hospital': [35006, 35006], 'patientID': [0003, 0004], 'results': [1,0]}]
 #Adult Patients(25-30yrs)


# Hospital35007 Adults Test results
diabetes_35007 = \
[{'hospital': [35007, 35007], 'patientID': [0001, 0002], 'results': [0,1]},\
#Adult Patients(18-25yrs)
{'hospital': [35007, 35007], 'patientID': [0003, 0004], 'results': [1,0]}]
 #Adult Patients(25-30yrs)

First, your function to convert lists of dictionaries to DataFrame can be simplified:

def resultDF(test_results):
    return pd.concat([pd.DataFrame(res) for res in test_results]).reset_index(drop = True)

Then, what I am suggesting is to group all the results into a dictionary and convert all them to DF in one go:

test_dict = {35006 : diabetes_35006,
             35007 : diabetes_35007}

res_dict = {key: resultDF(el) for key, el in test_dict.iteritems()}  

So that you have:

res_dict[35006]
Out[64]: 
   hospital  patientID  results
0     35006          1        0
1     35006          2        1
2     35006          3        1
3     35006          4        0

and:

   hospital  patientID  results
0     35007          1        0
1     35007          2        1
2     35007          3        1
3     35007          4        0