计算嵌套字典中列表的均值

Question

I have a nested dictionary, that I transformed in a pickle file. The pickle file can be found here . To open the pickle file is just like thar:

import pickle
resultados_acoes_testvar = pickle.load(open('map_results_modelo_acoes_variandotest.pickle', 'rb'))

The file is a dictionary, with that structure:

{'amat': {'Test_Size_100': {'raw_0': array([1.39838652e+02, 1.42292998e+02, 1.45314363e+02, 1.49162546e+02....)]}}}

Where "amat" is the name of the dataset(it has 9 datasets in the dict), test_size is the length of my prediction(prediction of a time serie), and raw is the model(it has 6 models in the dict) and the _0 is the time that I run. I run each model 10 times(0 to 9).

I would like to to get one time serie and for test_size and each model, with the mean of the nines times that I run each model.

I'm trying to do that way:

resultado = {}

lista_modelos = ['raw','difference', 'logaritmica', 'box_cox', 'mas', 'pct']

for acao in resultados_acoes_testvar_transformadas.keys():

resultado[acao] = {}

for testsize in resultados_acoes_testvar_transformadas[acao].keys():
    
    resultado[acao][testsize] = {}
    
    for values in resultados_acoes_testvar_transformadas[acao][testsize].keys():
        
        for prefix in lista_modelos:
            
            resultado[acao][testsize][prefix] = []
            

            for a, b, c, d, e, f, g, h, i, j in zip(values[prefix + '_0'],values[prefix+'_1'],values[prefix+'_2'],values[prefix+'_3'],values[prefix+'_4'],values[prefix+'_5'],values[prefix+'_6'],values[prefix+'_7'],values[prefix+'_8'],values[prefix+'_9']):
                           
                mean = float((a+b+c+d+e+f+g+h+i+j)/10)
                resultado[acao][testsize][prefix].append(mean)     

But I'm getting a error:

    TypeError                                 Traceback (most recent call last)
<ipython-input-59-80ef15c9251e> in <module>
     19 
     20 
---> 21                 for a, b, c, d, e, f, g, h, i, j in zip(values[prefix + '_0'],values[prefix+'_1'],values[prefix+'_2'],values[prefix+'_3'],values[prefix+'_4'],values[prefix+'_5'],values[prefix+'_6'],values[prefix+'_7'],values[prefix+'_8'],values[prefix+'_9']):
     22 
     23                     mean = float((a+b+c+d+e+f+g+h+i+j)/10)

TypeError: string indices must be integers

Thanks for your help.

Answer 1

Can you check if this works:

lista_modelos = ['raw','difference', 'logaritmica', 'box_cox', 'mas', 'pct']
for acao in resultados_acoes_testvar_transformadas.keys():
    resultado[acao] = {}
    for testsize in resultados_acoes_testvar_transformadas[acao].keys():
        resultado[acao][testsize] = {}
        for values in resultados_acoes_testvar_transformadas[acao][testsize].keys():
            for prefix in lista_modelos:
                resultado[acao][testsize][prefix] = []
                subd = resultados_acoes_testvar_transformadas[acao][testsize]  # <- HERE
                for a, b, c, d, e, f, g, h, i, j in zip(subd[prefix + '_0'],subd[prefix+'_1'],subd[prefix+'_2'],subd[prefix+'_3'],subd[prefix+'_4'],subd[prefix+'_5'],subd[prefix+'_6'],subd[prefix+'_7'],subd[prefix+'_8'],subd[prefix+'_9']):
                    mean = float((a+b+c+d+e+f+g+h+i+j)/10)
                    resultado[acao][testsize][prefix].append(mean)

Answer 2

尝试打印变量值，它可能是一个字符串