2026-04-20T08:48:39Z https://meral.edu.mm/oai

oai:meral.edu.mm:recid/00008776 2024-05-27T03:52:35Z 1582963436320 1582963436320:1582965742757

Data Mining with Emphasis on Exploratory Analysis (Aye Aye Win, 2015) Aye Aye Win Data mining is an analytical tool that is used in solving critical decision making problems by analyzing enormous amount of data in order to discover relationships and unknown patterns among variables in the data. This study focused on the investigation of the application of data mining techniques based on the tuberculosis (TB) diagnosis data set. The required data were organized from 659 TB suspected patients who came to the Union Tuberculosis Institute (UTI), Yangon during September and October 2013. This study attempted to predict whether a TB suspect has TB or not through the classification models by using decision tree method under the data mining techniques. The classification task with five different algorithms was made using decision tree method. It was found that the decision tree model of Algorithm I was found to be less accurate which used original data without preprocessing. The other four models which have performed preprocessing task revealed a better prediction having the same accuracy. Thus, this study proved that the decision tree method did not need the use of variable aggregation and feature reduction. The findings indicated that Active Specific Lung Lesion variable is the best predictor for making diagnosis about the present or absence of TB. The categorical value ‘Yes’ on Active Specific Lung Lesion is the most significant predictor of TB. Besides, the results obtained from decision tree method were compared with the results from logistic regression method. It was able to show that the accuracy of prediction for existence of TB disease or not is the same in two methods. It has also been observed that decision tree technique can provide classification rules which can identify the symptoms of TB. Therefore, decision tree method is found to be advantageous for the complex problems to make correct decisions according to the application used in this dissertation. Moreover, an alternative decision tree model was constructed without including X-ray result (Active Specific Lung Lesion variable). Even though the results from this was less accurate model using only patient’s symptoms, the rules of this model were useful for people who had not undergone medical check-up at clinics in order to the predict the present of TB. The classification rules provided by the decision tree model (without X-ray results) revealed that there is a better advantageous for the healthcare centers which have no X-ray machine since these rules can be used to make the efficient prediction for diagnosis. By using these rules (in Appendix D), the field workers should encourage the patient who has high likelihood of TB positive to go to the nearest healthcare center where X-ray machine, advanced technologies for diagnosis and expert technicians has. 2015-06-01 https://meral.edu.mm/records/8776