MasterThesis/tests.py at master · AlessandroPerini/MasterThesis · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
import postProcessing
import preProcessing
import utility
import resultsVisualization
from fileWriter import FileWriter
import pandas as pd
import oneHotEncoding
from sklearn import tree
import time


def test_a_priori_free_tuples_selection(x, y, tuples_selection_mode, max_depth=None):
    classifier = tree.DecisionTreeClassifier(max_depth=max_depth)
    x, y, list_y, computation_time = preProcessing.starter(x, y, tuples_selection_mode)
    #x.drop(axis=1, columns=['age', 'Male', 'Female'], inplace=True)
    classifier.fit(x, list_y)
    resultsVisualization.tree_printer(classifier, x, tuples_selection_mode)
    explanations = resultsVisualization.path_finder(classifier, x, list_y)
    resultsVisualization.print_explanations_to_terminal(explanations)
    purity, tree_height, n_imp_nodes = utility.tree_features_calculator(classifier, x, list_y)
    resultsVisualization.features_to_terminal(tuples_selection_mode, computation_time, purity, tree_height, n_imp_nodes)
    return explanations, computation_time, purity, tree_height, n_imp_nodes


def test_all_free_tuples_selection(x,y, max_depth=None):
    """
    :param x: dataframe_x
    :param y: dataframe_y
    :return: fitted classifier + prints the tree
    """
    classifier = tree.DecisionTreeClassifier(max_depth=max_depth)
    y = utility.transform_y_to_all_results(x, y)
    y = oneHotEncoding.encoder(y, x)
    x = oneHotEncoding.encoder(x, x)
    x, y, list_y = utility.y_creator(x, y)
    #x.drop(axis=1, columns=['age', 'Male', 'Female'], inplace=True)
    #y.drop(axis=1, columns=['age', 'Male', 'Female'], inplace=True)
    classifier.fit(x, list_y)
    resultsVisualization.tree_printer(classifier, x)
    return classifier, x, y, list_y


def test_a_posteriori_free_tuples_selection(x, y, tuples_selection_mode, max_depth=None):
    classifier, x, y, list_y = test_all_free_tuples_selection(x, y, max_depth)
    applied = classifier.apply(x)
    important_nodes = list()
    for elem in range(len(list_y)):
        if list_y[elem] != 0:
            important_nodes.append(applied[elem])

    print('Important nodes BEFORE:')
    print(important_nodes)
    start = time.time()
    if tuples_selection_mode == 'min':
        y, list_y = postProcessing.min_altitude_first(x, y, list_y, classifier)
    else:
        tuples_selection_mode = 'most'
        y, list_y = postProcessing.most_important_node_first(classifier, x, y, list_y)

    end = time.time()
    computation_time = end - start
    print('\nTime needed for \'post processing\': ' + str(end - start) + ' seconds\n')
    print(y)
    print(list_y)
    important_nodes = utility.important_nodes_generator(classifier, x, list_y)
    resultsVisualization.tree_printer(classifier, x, tuples_selection_mode, important_nodes)
    print('Important nodes AFTER:')
    print(important_nodes)
    explanations = resultsVisualization.path_finder(classifier, x, list_y)
    resultsVisualization.print_explanations_to_terminal(explanations)
    purity, tree_height, n_imp_nodes = utility.tree_features_calculator(classifier, x, list_y)
    resultsVisualization.features_to_terminal(tuples_selection_mode, computation_time, purity, tree_height, n_imp_nodes)
    return explanations, computation_time, purity, tree_height, n_imp_nodes


def test_all_free_tuples_combinations(x, y):

    y = utility.transform_y_to_all_results(x, y)
    results = list()
    results.append(pd.DataFrame())
    n_freesets = y.tupleset.unique().tolist()
    for set in n_freesets:
        result_temp = list()
        y_rows = y[y.tupleset == set]
        temp_results = len(results)
        for num_of_df in range(temp_results):
            for row in range(y_rows.shape[0]):
                k = pd.concat([results[num_of_df], y_rows.iloc[[row]]], axis=0)
                result_temp.append(k)

        results = result_temp

    print(results)


def test_all(x, y, query_x, query_y, max_depth):

    explanations_list = list()
    times_list = list()
    purities_list = list()
    heights_list = list()
    numbers_imp_nodes_list = list()

    expl1, time1, purity1, height1, n_imp_nodes1 = test_a_priori_free_tuples_selection(x, y, 'r', max_depth)
    expl2, time2, purity2, height2, n_imp_nodes2 = test_a_priori_free_tuples_selection(x, y, 'c', max_depth)
    expl3, time3, purity3, height3, n_imp_nodes3 = test_a_posteriori_free_tuples_selection(x, y, 'min', max_depth)
    expl4, time4, purity4, height4, n_imp_nodes4 = test_a_posteriori_free_tuples_selection(x, y, 'most', max_depth)

    explanations_list.extend((expl1, expl2, expl3, expl4))
    times_list.extend((time1, time2, time3, time4))
    purities_list.extend((purity1, purity2, purity3, purity4))
    heights_list.extend((height1, height2, height3, height4))
    numbers_imp_nodes_list.extend((n_imp_nodes1, n_imp_nodes2, n_imp_nodes3, n_imp_nodes4))
    weighted_parameters = list()

    for i in range(len(heights_list)):
        weighted_parameters.append((heights_list[i]+numbers_imp_nodes_list[i]*1.2) / (purities_list[i]/100))

    best_parameter = min(weighted_parameters)
    index = 0
    for weighted_parameter in weighted_parameters:
        if weighted_parameter == best_parameter:
            best_method = index
        index += 1
    file = FileWriter(query_x, query_y, y)
    file.times_writer(times_list)
    file.explanations_writer(explanations_list)
    file.purity_writer(purities_list)
    file.heights_writer(heights_list)
    file.number_important_nodes_writer(numbers_imp_nodes_list)
    file.best_method_writer(best_method, best_parameter)