Calculating Feature Weights in Naive Bayes with Kullback-Leibler Measure

Lee, Changhwan; Gutiérrez, Fernando; Dou, Dejing

doi:10.1109/icdm.2011.29

Cited by 85 publications

(69 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This might lead to discover new subsets of features with a high impact on the classification performance. 6,7,8,9,11,13,14,15,20,22,24,25,26,27,28 0.048 0.055 …”

Section: Discussionmentioning

confidence: 99%

“…There are some research attempts in the literature that investigated the effect of feature selection on improving the accuracy of classification techniques [11,10,14]. As for phishing detection, A. Bergholz et el.…”

Section: Related Workmentioning

confidence: 99%

“…Two feature combinations achieve the best overall accuracy of 94.87% for SVM classifier. The first comprises of the following 16 features: 2, 6,7,8,9,11,13,14,15,19,20,24,25,26,27,28. The second comprises of the same 16 features in the first combination except replacing the Redirect feature with popUpWidnow.…”

Section: Evaluation With Omitting Redundant Featuresmentioning

confidence: 99%

“…Table 5 shows the most effective subset of features selected by wrapper method that results in the best performance for each classifier. The feature combination that achieves the best overall accuracy of 94.67 % for decision tree classifier comprises of the following 12 features (listed by their index in Table 1): 2, 6,7,8,9,11,14,15,24,25,26,30. The feature combination that achieves the best overall accuracy of 93.93% for Naive Bayes classifier compromises of the following 10 features: 2, 6, 7, 8, 9, 14, 16, 26, 28, 29. The feature combination that achieves the best overall accuracy of 94.54% for SVM classifier compromises of the following 18 features: 1, 2, 3, 5, 6, 7, 8, 10, 13, 14, 16, 19, 21, 22, 26, 27, 28, 30. …”

Section: Evaluation With Omitting Redundant Featuresmentioning

confidence: 99%

“…The feature combination that achieves the best overall accuracy of 95.40% for decision tree classifier comprises of the following 14 features (listed by their index in Table 1) : 6, 7, 8, 9, 10, 13, 14, 15, 20, 24, 25, 26, 27, 28. The feature combination that achieves the best overall accuracy of 94.055% for Naive Bayes classifier comprises of the following 16 features: 2, 3,6,7,8,9,11,13,14,15,20,24,25,26,27,28. Two feature combinations achieve the best overall accuracy of 94.87% for SVM classifier.…”

Section: Evaluation With Omitting Redundant Featuresmentioning

confidence: 99%

See 4 more Smart Citations

On Determining the Most Effective Subset of Features for Detecting Phishing Websites

Hassan¹

2015

IJCA

View full text Add to dashboard Cite

Phishing websites are a form of mimicking the legitimate ones for the purpose of stealing user 's confidential information such as usernames, passwords and credit card information. Recently machine learning and data mining techniques have been a promising approach for detection of phishing websites by distinguishing between phishing and legitimate ones. The detection process in this approach is preceded by extracting various features from a website dataset to train the classifier to correctly identify phishing sites. However, not all extracted features are effective in classification or equivalent in their contribution to its performance. In this paper, we investigate the effect of feature selection on the performance of classification for predicting phishing sites. We evaluate various machine learning algorithms using a number of feature subsets selected from an extracted feature set by various feature selection techniques in order to determine the most effective subset of features that results in best classification performance. Empirical results shows that using our new proposed methodology for selecting features by removing redundant ones that equally contribute to the classification accuracy, the decision tree classifier achieves the best performance with an overall accuracy of 95.40%, false positive rate (FPR) of 0.046 and false negative rate (FNR) of 0.065.

show abstract

“…This might lead to discover new subsets of features with a high impact on the classification performance. 6,7,8,9,11,13,14,15,20,22,24,25,26,27,28 0.048 0.055 …”

Section: Discussionmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%