lyzq/malicious-url-app
0
1+
2 m6�i� � �� � ^ RI Ht ^ RIt^ RIHt ^ RIHt ^ RIt ^ RI3t ^ RI t
^ RIHtHtHtHtHtHtHtHtHtHt RR R lltR R ltR R4 ltR R lt R
R lt R R lt R R lt R R lt R# )� )� annotationsN)�Path)5�ConfusionMatrixDisplay�average_precision_score�balanced_accuracy_score�classification_report�confusion_matrix�f1_score�precision_recall_curve�precision_score� recall_score�
roc_auc_scorec 6 �, � V ^8� d QhRRRRRRRRRR /# )7� �y_true� pd.Series� dataset_name�str�8model_name� threshold�float�return�dict� )�formats "�src\reporting.py� __annotate__r s< � � � � �� �� � �9 � � 10�
� c � � W8� P \ 4 pR VRVRVR\ \ W^ R7 4 R\ \ W^ R7 4 R\ \ W^ R7 4 R\ \
W4 4 R\ \ W4 4 R \ \ W4 4 R11\ W4 P 4 R \ WR ^ R
7 / # )�dataset�modelr � precision)�
zero_division�recall�f1�balanced_accuracy�roc_auc�average_precisionr r T)� output_dictr# ) �astype�intr r r
r12 r r r r �tolistr )r �y_scorer r r �y_preds &&&&& r �evaluate_predictionsr/ s� � � �"�
*�
*�3�
/�F��<����Y��U�?�6��K�L��%� �V�1�E�F� �e�H�V�1�=�>��U�#:�6�#J� K��5� �v�7�8��U�#:�6�#K� L�� ,�V� <� C� C� E� �!6�v�SW�gh�!i� � r c �$ � V ^8� d QhRRRRRR/# )r �metrics�13list[dict]�pathr r �Noner )r s "r r r / s& � � E� E�*� E�D� E�T� Er c �X � VP \ P ! V ^R7 RR7 R# )r )�indentzutf-8)�encodingN)�14write_text�json�dumps)r1 r3 s &&r � save_metricsr; / s � ��O�O�D�J�J�w�q�1�G�O�Dr c �$ � V ^8� d QhRRRRRR/# )r r r �target_precisionr r r )r s "r r r 3 s# � � � �)� �PU� �Z_� r c � � \ W4 w r4pR pRp\ VRR VRR V4 F, w r�p15W�8� g K W�8� g K \ V164 p\ V 4 pK. V# )� �?Ng ������)r �zipr ) r r- r= r" r$ �17thresholds�selected� best_recall�prec�recr s &&& r � select_threshold_by_precisionrG 3 se � �$:�6�$K�!�I�z��H��K� #�I�c�r�N�F�3�B�K�� L� ��9� � #��(9� �Y�'�H� ��*�K� !M� �Or c �$ � V ^8� d QhRRRRRR/# )r r1 r2 r3 r r � pd.DataFramer )r s "r r r > s! � � � �18� �$� �<� r c � � \ P ! V 4 p. ROpW#, P R R.RR.R7 pVP VRR7 V# )r r% TF��by� ascending)�index)r r! r" r$ r% r& r' r( )�pd� DataFrame� sort_values�to_csv)r1 r3 �frame�columns�summarys && r �save_summary_tablerV > sO � � �L�L�� !�E� �G� �n�(�(�Y��,=�$�PU� �(�W�G� �N�N�4�u�N�%� �Nr c �$ � V ^8� d QhRRRRRR/# )r rU rI � output_pathr r r4 r )r s "r r r O s! � � � �l� �� �$� r c �� � \ P ! RR7 V P RR.. RORRR7 p\ P ! VRRRRR7 \ P19 ! ^ R 4 \ P ! R204 \ P ! R 4 \ P ! R 4 \ P ! 4 \ P ! V^�R
7 \ P ! 4 R# )�21 ��figsizer r! �metric�score)�id_vars�22value_vars�var_name�23value_name�Set2)�data�x�y�hue�paletteg�������?z$Model Comparison Across Core Metrics�Score� ��dpiN)rZ � )r% r' r( ) �plt�figure�melt�sns�barplot�ylim�title�ylabel�xlabel� tight_layout�savefig�close)rU rX �melteds && r �plot_metric_barsr{ O s� � ��J�J�w� �
�\�\��G�$�9��� � �F� �K�K�V�x�7���P��H�H�Q����I�I�4�5��J�J�w���J�J�r�N������K�K� ��%��I�I�Kr c �$ � V ^8� d QhRRRRRR/# )r r1 r2 rX r r r4 r )r s "r r r a s! � � 24� 25�Z� 26�d� 27�t� 28r c �� � \ P ! ^^R
R7 w r#VP 4 p\ W04 Fh w rE\ P29 ! VR, 4 p\
VRR.R7 pVP VRR7 VP VR, R VR30, 24 Kj \ P ! 4 \ P ! V^�R 7 \ P ! 4 R # )r r[ r �Benign� Malicious)r �display_labelsF)�ax�colorbarr! z | r rk N)rZ � ) rn �subplots�flattenrA �np�arrayr �plot� set_titlerw rx ry )r1 rX �fig�axesr� r] �matrix�displays && r �plot_confusion_matricesr� a s� � �� � �Q��7�3�I�C� �<�<�>�D��$�(�31�����&�!3�4�5��(�&�RZ�\g�Qh�i��� � ��U� �+�32� � ��w��(��F�9�,=�+>�?�@� )�33 �����K�K� ��%��I�I�Kr c �$ � V ^8� d QhRRRRRR/# )r � top_featuresr rX r r r4 r )r s "r r r n s! � � � �t� �$� �4� r c � � \ P ! V R , 4 P RRR7 p\ P ! R R7 \ P34 ! VR, VR, RR7 \ P ! R4 \ P ! R 4 \ P ! 4 \ P ! V^�R357 \ P ! 4 R # )
� malicious�weightTrK r[ �feature�#2457a5)�colorz4Top Character n-grams Associated with Malicious URLszCoefficient Weightrk N)r� rm ) rO rP rQ rn ro �barhrt rv rw rx ry )r� rX rS s && r �plot_logistic_ngramsr� n s� � � �L�L� �k�2� 3� ?� ?�8�W[� ?� \�E��J�J�v� ��H�H�U�9�
�u�X��i�@��I�I�D�E��J�J�#�$������K�K� ��%��I�I�Kr c �$ � V ^8� d QhRRRRRR/# )r �historyr2 rX r r r4 r )r s "r r r y s! � � � �:� �D� �T� r c �p � \ P ! V 4 p\ P ! RR7 \ P ! VR, VR, RRRR7 \ P ! VR, VR, RR R36R7 \ P37 ! R 4 \ P ! R 4 \ P ! R
4 \ P ! VR, 4 \ P ! 4 \ P ! 4 \ P ! V^�R7 \ P ! 4 R# )r� r[ �epoch�38train_loss�oz39Train lossr� )�marker�labelr� �val_losszValidation lossz#d26a2ez Character CNN Training Curve�Epoch�Lossrk N)r� � )
rO rP rn ro r� rt rv ru �xticks�legendrw rx ry )r� rX rS s && r �plot_training_historyr� y s� � � �L�L�� !�E��J�J�v� ��H�H�U�7�^�U�<� 0��L�Xa�b��H�H�U�7�^�U�:� .�s�BS�[d�e��I�I�,�-��J�J�w���J�J�v���J�J�u�W�~� ��J�J�L������K�K� ��%��I�I�Kr )r? )!�40__future__r r9 �pathlibr �matplotlib.pyplot�pyplotrn �numpyr� �pandasrO �seabornrq �sklearn.metricsr r r r r r41 r r r
r r/ r; rG rV r{ r� r� r� r r r �<module>r� sR �� "� � � � � � � � � � �.E���"�$42�� r