A new hybrid model of dengue incidence rate using negative binomial generalised additive model and fuzzy c-means model: a case study in Selangor

Mohamad, Nazeera (2018) A new hybrid model of dengue incidence rate using negative binomial generalised additive model and fuzzy c-means model: a case study in Selangor. Masters thesis, Universiti Tun Hussein Onn Malaysia.

Preview	Text 24p NAZEERA MOHAMAD.pdf Download (18MB) | Preview
	Text (Copyright Declaration) NAZEERA MOHAMAD COPYRIGHT DECLARATION.pdf Restricted to Repository staff only Download (655kB) | Request a copy
	Text (Full Text) NAZEERA MOHAMAD WATERMARK.pdf Restricted to Registered users only Download (66MB) | Request a copy

Abstract

Dengue is one of the top reason for illness and mortality in the world with beyond one­third of the world's population living in the risk areas of dengue infection. In this study, there are five stages to achieve the research objectives. Firstly, the verification of predetem1ined variables. Secondly, the identification of new datasets after clustered by district and Fuzzy C-Means Model (FCM). Thirdly, the development of models using the existing dataset and the new datasets which clustered by the two different clustering categories. Then, to assess the models developed by using three measurement methods which are deviance (D), Akaike Jnfonnation Criteria (AIC) and Bayesian Infonnation Criteria (BIC} Lastly, the validation of model developed by comparing the value of D, AIC and BIC between the existing model and the new models developed which used the new datasets. There are two different clustering techniques applied which are clustering the data by district and by FCM. This study proposed a new modelling hybrid framework by using two statistical models which are FCM and negative binomial Generalised Additive Model (GAM). This study successfully presents the significant difference in the climatic and non-climatic factors that influenced dengue incidence rate (DIR) in Selangor, Malaysia. Results show that the climatic factors such as rainfall with current month up to 3 months and number of rainy days with current month up to lag 3 months are significant to DIR. Besides, the interaction between rainfall and number of rainy days also shows strong positive relationship to DIR. Meanwhile, non-climatic vaiiables such as population density, number of locality and lag DIR from I month until 3 months also show significant relationship towards DIR For both clustering techniques, there are two clusters fonned and there are four new models developed in this study. After comparing the values of D, AIC ai1d BIC between the existing model and the new models, this study concluded that four new models recorded lower values compared to the existing model. Therefore, the four new models are selected to present the dengue incidence in Selangor.

Item Type:	Thesis (Masters)
Subjects:	R Medicine > RC Internal medicine R Medicine > RC Internal medicine > RC109-216 Infectious and parasitic diseases
Divisions:	Faculty of Applied Science and Technology > Department of Physics and Chemistry
Depositing User:	Mrs. Sabarina Che Mat
Date Deposited:	15 Aug 2022 03:07
Last Modified:	15 Aug 2022 03:07
URI:	http://eprints.uthm.edu.my/id/eprint/7517

Actions (login required)

View Item