Data Mining of Access to Tetanus Toxoid Immunization Among Women of Childbearing Age in Ethiopia

Abstract

Tetanus toxoid (TT) vaccine is given to women of childbearing age to prevent neonatal tetanus and maternal mortality attributed to tetanus. Globally, tetanus is responsible for 5% of maternal deaths and 14% of neonatal deaths annually. Data mining is the process of discovering interesting patterns and knowledge from large amounts of data. Thus, the aim of this study was to identify the best classifier, and to predict the pattern from the TT data set using the data mining algorithms technique. The data for this study were the Tetanus Toxoid data set from the Ethiopian Demographic and Health Survey (EDHS) 2011, and analyzed using the Knowledge discovery process of Selection, Processing, Transforming, mining, and interpretation. The WEKA 3.6.1 tool was used for classification, clustering, association and attribute selection. The accuracy rate of the classifiers on training data is relatively higher than on test data and the multilayer perceptron is the best classifier in our data set on Tetanus toxoid. In the cross-validation with 10 folds, correctly classified best are by naive Bayesian 63.30% and the least accurate were by k-nearest neighbor 60.52%. Single data instance test using Naive Bayesian was done by creating test 1, test 2, test 3, and test 4 data test instance, three of them are correctly predicted but one of them incorrectly classified. The maximum confidence attained in the general association is 0.98. But, in the class attribute, it is 0.72. The literacy status of the mother has high information gain with the value 0.046. As a conclusion, the best algorithm based on the TT vaccination data is multilayer perceptron classifier with an accuracy of 67.28% and the total time taken to build the model is at 0.01 seconds. Multilayer perceptron classifier has the lowest average error at 32.72% compared to others. These results suggest that among the machine learning algorithm tested, multilayer perceptron classifier has the potential to significantly improve the conventional classification methods for use in EDHS data of Tetanus toxoid.