Addis Ababa University Libraries Electronic Thesis and Dissertations: AAU-ETD! >
School of Information Science and Computer Science >
Thesis - Health Informatics >
Please use this identifier to cite or link to this item:
|Title: ||PREDICTING LOW BIRTH WEIGHT USING DATA MINING TECHNIQUES ON ETHIOPIA DEMOGRAPHIC AND HEALTH SURVEY DATA SETS|
|Authors: ||BISET, DESALEGN|
|Advisors: ||Dr. Dereje Teferi|
Dr. Mitike Molla
|Keywords: ||health informatics|
|Copyright: ||Jun-2011 |
|Date Added: ||31-Jul-2012 |
|Abstract: ||Low birth weight is one of the critical issues in Ethiopia that causes many babies short- term and long-term health consequences and tend to have higher mortality and morbidity. DHS Ethiopia report shows that the percentage of low birth weight babies has increased from 8 percent in 2000 to 14 percent in 2005. The percentage of babies assessed by mothers as being very small at birth has increased over the same period from 6 percent to 21 percent
Low birth weight is a reasonable well-defined problem caused by many factors that are potentially modifiable and the costs of preventing them are well within reach, even in poor countries like Ethiopia. Therefore, it is very important to predict LBW in various communities in the country in order to come up with feasible intervention strategies to minimize the problem. Data mining techniques is a good tool to explore hidden knowledge from huge data set. The goal of this study is to predict low birth weight using EDHS 2005 (Ethiopia Demographic Health Survey) data set by applying of data mining technology. This study tried to build a model using data mining technique addressing the factors associated with low birth weight.
To this end, data was collected from Measure DHS Ethiopia. The methodology applied in this research was CRISP-DM, which contains six major phases: business understanding, data understanding, and data preparation, model building, evaluation and deployment. A total of 9861 records were used for the experiments. Some attributes of numeric values are discretized using ten-bin discretization implemented in Weka. Besides, missing values are also handling by data imputation technique, which replaces all missing values with the modes for nominal and categorical and means for numerical values from the training instances.
The selected data mining techniques for predicting low birth weight was classification. J48 decision tree classifier and PART rule induction algorithms were selected for experiments.
Several models were built implementing the J48 decision tree classifier algorithm and PART rule induction algorithms. These experiments has been done using pruning with all and reduced attributes, by giving J48 classifiers parameters in different values. The researcher compare the classification performance of the decision trees with tree prunning and without tree prunning, and found that tree prunning can significantly improve decision tree’s classification performance.
In general, the results from this study were encouraging; it can be used as decision support aid for health practitioner. The extracted rules in both the algorithms are very effective for the prediction of low birth weight. It is possible to observe, from both algorithms that the attributes such as antenatal visits during pregnancy (antenatal care for pregnancy), mother’s educational level, and marital status, Iodine contents in salt, region, and age of mother, numbers of birth order and wealth index as well as place of residence are the most determinant factors to predict low birth weight.|
|Description: ||A Thesis Submitted to the School of Graduate Studies of Addis Ababa University in Partial Fulfillment of the Requirements for the Degree of Master of Science in Health Informatics|
|Appears in:||Thesis - Health Informatics|
Items in the AAUL Digital Library are protected by copyright, with all rights reserved, unless otherwise indicated.