Student-Coding-Behavior/train_final_model.py at main · DSAatUSU/Student-Coding-Behavior · 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
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
from sklearn.ensemble import IsolationForest, RandomForestClassifier
from sklearn.neighbors import LocalOutlierFactor
from sklearn.svm import OneClassSVM
import numpy as np
import glob
import pickle
from IPython.display import clear_output
from sklearn.metrics import classification_report, f1_score, confusion_matrix
# Assuming you have X_train and y_train

if __name__ == "__main__":
    with open("./data/final_model_dataset.pkl", "rb") as fp:   # Unpickling
        temp_dataset = pickle.load(fp)

    with open(glob.glob("./data/RandomForestClassifier*.pkl")[0], "rb") as file:
        rf_clf = pickle.load(file)

    with open(glob.glob("./data/IsolationForest*.pkl")[0], "rb") as file:
        models = pickle.load(file)

    X_train_main, y_train_main, X_test_main, y_test_main = temp_dataset


    # Create dictionaries to store models for each algorithm
    # models = {
    #     'IsolationForest': {},
    #     # 'LocalOutlierFactor': {},
    #     #'OneClassSVM': {}
    # }

    # Fit models for each class using different algorithms
    # for class_label in np.unique(y_train_main):
    #     print(f"Fitting models for class {class_label}")
    #     indexes = np.where(np.reshape(y_train_main, (-1))==class_label)[0]
    #     X_class = X_train_main[indexes]

    #     # Isolation Forest
    #     iforest = IsolationForest(contamination='auto',n_estimators=5, random_state=42, bootstrap=True, n_jobs=-1)
    #     iforest.fit(X_class)
    #     models['IsolationForest'][class_label] = iforest

        # # Local Outlier Factor
        # lof = LocalOutlierFactor(contamination='auto', n_neighbors=5, novelty=True)
        # lof.fit(X_class)
        # models['LocalOutlierFactor'][class_label] = lof

        # # One-Class SVM
        # ocsvm = OneClassSVM(nu=0.02)
        # ocsvm.fit(X_class)
        # models['OneClassSVM'][class_label] = ocsvm


    num_classes = 3
    itteration = 0
    prev_macro_f1 = 0
    prev_test_class_report = ""
    max_itter = 1000
    prev_random_state = 0
    prev_conf_mat = []


    while itteration < max_itter:
        # Fit models for each class using different algorithms
        # Create dictionaries to store models for each algorithm
        random_state = np.random.randint(0, 1997)
        new_rf_clf = RandomForestClassifier(
            criterion= 'entropy', max_depth= 10, max_features= 'sqrt', min_samples_leaf= 5, min_samples_split= 5, n_estimators= 10, class_weight = {0:.5,1:.3,2:.2}, random_state=random_state
        )
        new_rf_clf.fit(X_train_main, y_train_main.ravel())

        models = {
            'IsolationForest': {},
            # 'LocalOutlierFactor': {},
            # 'OneClassSVM': {}
        }
        for class_label in [0, 1, 2]:
            #print(f"Fitting models for class {class_label}")
            indexes = np.where(np.reshape(y_train_main, (-1))==class_label)[0]
            X_class = X_train_main[indexes]

            # Isolation Forest
            iforest = IsolationForest(contamination='auto',n_estimators=20, random_state=random_state, bootstrap=True, n_jobs=-1)
            iforest.fit(X_class)
            models['IsolationForest'][class_label] = iforest

            # # Local Outlier Factor
            # lof = LocalOutlierFactor(contamination='auto', n_neighbors=5, novelty=True)
            # lof.fit(X_class)
            # models['LocalOutlierFactor'][class_label] = lof

            # # One-Class SVM
            # ocsvm = OneClassSVM(nu=0.02)
            # ocsvm.fit(X_class)
            # models['OneClassSVM'][class_label] = ocsvm
        clear_output(wait=True)
        print("Progress: ", end="")
        print(f"*"*int(((itteration)/max_itter)*100)+ "-"*(int(((max_itter-itteration)/max_itter)*100)-100) + f" {((itteration)/max_itter)*100:.2f}%")
        all_predictions = []
        for test_index in range(len(X_test_main)):
            example = X_test_main[test_index]
            rf_pred_prob = new_rf_clf.predict_proba([example])[0]
            average_scores = {}
            for class_label in [0, 1, 2]:
                scores_sum = 0
                num_models = 0
                for algorithm_models in models.values():
                    model = algorithm_models[class_label]
                    score = model.score_samples([example])[0]
                    scores_sum += score
                    num_models += 1
                average_score = scores_sum / num_models
                average_scores[class_label] = average_score
            # Normalize average scores to be between 0 and 1
            #average_scores = {k: (v - min(average_scores.values())) / (max(average_scores.values()) - min(average_scores.values())) for k, v in average_scores.items()}
            if np.argmax(rf_pred_prob) == 1:
                predicted_class = np.argmax(rf_pred_prob)
            else:
                score = list(average_scores.values())
                predicted_class = np.argmax(score)
            all_predictions.append(predicted_class)

        test_clf_report = classification_report(y_test_main.ravel(), all_predictions, zero_division=1)
        test_macro_f1 = f1_score(y_test_main.ravel(), all_predictions, average='macro')
        conf_mat = confusion_matrix(y_test_main.ravel(), all_predictions)

        if prev_macro_f1 < test_macro_f1:
            prev_random_state = random_state
            prev_macro_f1 = test_macro_f1
            prev_test_class_report = test_clf_report
            prev_conf_mat = conf_mat

        itteration += 1
    print(f"Test Class Report {prev_random_state}:\n {prev_test_class_report}")
    print(f"Confusion Matrix {prev_random_state}:\n {prev_conf_mat}")


    # while itteration < max_itter:
    #     clear_output(wait=True)
    #     print("Progress: ", end="")
    #     print("*"*itteration+ "__"*(max_itter-itteration) + f" {itteration}%")
    #     all_predictions = []
    #     for test_index in range(len(X_test_main)):
    #         example = X_test_main[test_index]
    #         rf_pred_prob = rf_clf.predict_proba([example])[0]
    #         average_scores = {}
    #         for class_label in range(num_classes):
    #             scores_sum = 0
    #             num_models = 0
    #             for algorithm_models in models.values():
    #                 model = algorithm_models[class_label]
    #                 score = model.score_samples([example])[0]
    #                 scores_sum += score
    #                 num_models += 1
    #             average_score = scores_sum / num_models
    #             average_scores[class_label] = average_score
    #         # Normalize average scores to be between 0 and 1
    #         #average_scores = {k: (v - min(average_scores.values())) / (max(average_scores.values()) - min(average_scores.values())) for k, v in average_scores.items()}
    #         if np.argmax(rf_pred_prob) == 1:
    #             predicted_class = np.argmax(rf_pred_prob)
    #         else:
    #             score = list(average_scores.values())
    #             predicted_class = np.argmax(score)
    #         all_predictions.append(predicted_class)

    #     test_clf_report = classification_report(y_test_main.ravel(), all_predictions, zero_division=1)
    #     test_macro_f1 = f1_score(y_test_main.ravel(), all_predictions, average='macro')

    #     if prev_macro_f1 < test_macro_f1:
    #         prev_macro_f1 = test_macro_f1
    #         prev_test_class_report = test_clf_report

    #     itteration += 1
    # print(f"Test Class Report:\n {prev_test_class_report}")