1. UKM Learning and Research Repository
2. Master Thesis / Tesis Master
3. Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Title:	Feature selection algorithms for Dengue outbreak detection mode
Authors:	Abuhamad Husam I. S. (P63049)
Supervisor:	Azuraliza Abu Bakar, Prof. Dr.
Keywords:	Dengue outbreak Mosquito borne disease Genetic algorithms
Issue Date:	6-Mar-2013
Description:	Dengue outbreak is considered as one of the most common mosquito borne disease worldwide specifically in Malaysia. According to World Health Organization (WHO), dengue fever (DF) and dengue hemorrhagic fever (DHF) are considered as public health related aspect in Malaysia and other countries. Dengue outbreak is defined as occurrence of more than one case in the same locality, where the date of onset between the cases less than 14 days. The outbreak is clear when no new case has been reported within 14 days. Therefore, it is important to avoid a rapid spread of dengue outbreak by having the knowledge that enables to predict the next outbreak to occur. Many previous researches have explored the use of data mining techniques to solve the problem. They used thousands of datasets with more than one hundred attributes collected from the health department. To date, with the average of 70%-80% accuracy prediction, the most related attributes or features that contribute to the dengue outbreak prediction accuracy have not yet explored. Thus, this study aims to apply a feature selection process for dengue dataset. Genetic algorithm (GA), particle swarm optimization (PSO) and rank search (RS) are chosen as the feature selection algorithms due to their success to perform this task and in wide range of application in medical domain. The original dengue datasets is obtained from the Public Health Department, Seremban, Negeri Sembilan which contains 6082 data and 21 attributes. In this study several additional attributes that are the weather data attributes are merged to the original dataset as suggested by the expert. Two phases are conducted. Firstly, Phase I investigates the selected features from the feature selection algorithms in terms of enhancing the predictive accuracy of dengue outbreak detection. Phase II focuses on the predictive modeling task of dengue outbreak problem using three classification algorithms namely J48 decision tree, DTNB naive Bayes and artificial neural network. The experimental results show that the PSO algorithm has selected the most related features in dengue dataset and this result has been verified by a group Public Health experts. PSO-J48 model has reached the highest accuracy among the other predictive modeling techniques. This study has revealed new set of features to represent dengue data and has enhanced predictive accuracy of dengue outbreak detection.,Master / Sarjana
Pages:	93
Call Number:	QA402.5.A2337 2013
Publisher:	UKM, Bangi
Appears in Collections:	Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.