In [1]:

    

import pandas as pd
df = pd.read_csv('queryset_CNN.csv')
print(df.shape)
print(df.dtypes)


    

    




(32, 3)
doc_id         int64
author_id      int64
authorList    object
dtype: object

In [2]:

    

# for index, row in df.iterrows():
# doc_id = row.doc_id
# import ast
# authorList = ast.literal_eval(row.authorList)
authorList = [11, 18, 80, 88, 64, 44, 91, 19, 97]

doc_id = 1

parameters = {
    'candidate': [2, 3, 4, 5, 6, 7, 8, 9],
    'samples': [320, 1600, 3200],
    'dropout': [0.0, 0.1, 0.2, 0.3, 0.4, 0.5],
    'dimensions': [50, 100, 200],
    'iterations': [10, 20, 40, 80, 120, 240],
    'cv': [320, 1600]#, 3200],
}

# print('doc_id: %3s actual author: %4s author_id: %s' % (str(doc_id), str(row.author_id), str(authorList)))

level = 'word'

for idxp, paralist in parameters.iteritems():
    if idxp == 'samples':
        for idxl, val in enumerate(paralist):
            
            authorList = [11, 18, 80, 88, 64, 44, 91, 19, 97]
            
            doc_id = 1

            candidate = 4

            test = idxp # change before run

            level = "word"

            iterations = 30

            dropout = 0.5

            samples = val

            dimensions = 200

            # loc = authorList.index(author_id)

            printstate = (("doc_id = %s, candidate = %s, ") % (str(doc_id), str(candidate)))
            printstate += (("dimensions = %s, samples = %s, ") % (str(dimensions), str(samples)))
            printstate += (("\niterations = %s, dropout = %s, test = %s") % (str(iterations), str(dropout), str(test)))

            print("Current test: %s" % (str(printstate)))

            import UpdateDB as db

            case = db.checkOldCNN(doc_id = doc_id, candidate = candidate, dimensions = dimensions,
                                  samples = samples,iterations = iterations, dropout = dropout, 
                                  test = test)

            if case == False:

                print("Running: %12s" % (str(printstate)))

                import StyloNeural as Stylo
                (labels_index, history, train_acc, val_acc, samples) = Stylo.getResults(
                    doc_id = doc_id, authorList = authorList[:candidate], 
                    level = level, glove = '../../glove/', dimensions = dimensions, 
                    samples = samples, nb_epoch = iterations, dropout = dropout, batch_size = 10 )

                # for key, auth in labels_index.iteritems():
                #     if auth == author_id:
                #         loc = key

                # ans = predY[loc]
                
                # count = 0
                
                # for train_acc, val_acc in zip(train_acc_list, val_acc_list):
                    
                import UpdateDB as db
                case = db.updateresultOldCNN(doc_id = doc_id, candidate = candidate, dimensions = dimensions,
                                             samples = samples, iterations = iterations, dropout = dropout,
                                             train_acc = train_acc, val_acc = val_acc, 
                                             test = str(test))# + '-fold-' + str(count)))
                #     count = count + 1

                # if case:
                #     ans = 0.0

                #     if (predY.tolist().index(max(predY)) == loc):
                #         ans = 1.0
                # 
                #     import UpdateDB as db
                #     case = db.updateresultOldCNN(doc_id = doc_id, candidate = candidate, dimensions = dimensions,
                #                                  samples = samples, iterations = iterations, dropout = dropout, 
                #                                  accuracy = ans, test = str(test + 'binary'))

                # else:
                #     print("Skipped: %12s" % (str(printstate)))
                
                del Stylo

                from keras import backend as K
                K.clear_session()

                import time
                time.sleep(10)

                from IPython.display import clear_output
                clear_output()

            else:
                print("Skipped: %12s" % (str(printstate)))


    

In [3]:

    

%tb


    

    




No traceback available to show.

In [ ]: