Classification of Stroke Opportunities with Neural Network and K-Nearest Neighbor Approaches

Abstract

Stroke is one of the deadly diseases. This is illustrated in stroke deaths in Indonesia which reached a death rate of 131.8 cases. Some of the things that cause a stroke to become a disease with the highest mortality rate are related to transitions in human life in 4 aspects, namely epidemiology, demography, technology, and economics, socio-culture. Of the many influencing aspects, one of the transition points of human life in the technological aspect can be an alternative solution and prevention. Aspects of technology with the utilization of data can be used as a preventive measure for stroke. One approach is to use data mining techniques, which can provide an initial picture regarding the chances of getting a stroke so that it can be used as an early warning for patients. With so many techniques in data mining, this study used a classification or grouping approach using 2 algorithms, namely K-Nearest Neighbor and one of the Neural Network groups, namely Multi-Layer Perceptron. This research will focus on finding the accuracy and best results of the two algorithms in classifying. The final result of this study is that the K-Nearest Neighbor algorithm has a better accuracy of 95% compared to the Multi-Layer Perceptron which produces an accuracy of 88%