Disjunctive naïve Bayesian classifier to enhance accuracy for dynamic prediction

Md Faisal Kabir, Hossain Ashik Mahmud Chowdury, Keshav Dahal, Alamgir Hossain, Chowdhury Mofizur Rahman

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    A novel supervised learning algorithm named disjunctive naïve Bayesian classifier is presented in this paper. In conventional naïve Bayesian classifier only one set of class conditional probabilities are calculated from the given data. However, in our proposed approach we divide the data using k-means clustering -and save the center of each cluster. We then train these k clusters in naïve Bayesian classifier. For a new case to classify we compute similarity with the previously obtained cluster centers and based on the best match, we use the appropriate cluster set conditional probability to predict the class. We tested our proposed model on a number of benchmark data and attained higher classification accuracy rates than conventional naïve Bayesian classifier.

    Original languageEnglish (US)
    Title of host publicationSKIMA 2010 - Proceedings of the 4th International Conference on Software, Knowledge, Information Management and Applications
    Subtitle of host publication"Towards Happiness and Sustainable Development"
    EditorsOuyporn Tonmukayakul, Manawin Songkroh, Pradorn Sureephong
    PublisherSKIMA
    Pages265-269
    Number of pages5
    ISBN (Electronic)9789746725569
    StatePublished - 2010
    Event4th International Conference on Software, Knowledge, Information Management and Applications: Towards Happiness and Sustainable Development, SKIMA 2010 - Paro, Bhutan
    Duration: Aug 25 2010Aug 27 2010

    Publication series

    NameSKIMA 2010 - Proceedings of the 4th International Conference on Software, Knowledge, Information Management and Applications: ''Towards Happiness and Sustainable Development''

    Conference

    Conference4th International Conference on Software, Knowledge, Information Management and Applications: Towards Happiness and Sustainable Development, SKIMA 2010
    CountryBhutan
    CityParo
    Period8/25/108/27/10

    All Science Journal Classification (ASJC) codes

    • Software

    Fingerprint Dive into the research topics of 'Disjunctive naïve Bayesian classifier to enhance accuracy for dynamic prediction'. Together they form a unique fingerprint.

    Cite this