Utilizing Machine Learning and Data Mining to Identify Diarrheal Diseases with the Decision Tree Method

In Indonesia, diarrheal disease is still a global health problem. The degree of morbidity and mortality is quite high, ranking 3rd after TB and pneumonia. In developed countries, even though there have been improvements in people’s health and economy, the incidence of infectious diarrhea remains high and is still a health problem. The development of increasingly complex science encourages humans to learn data that can be processed and produce new information and knowledge. In computer science, Machine learning (ML) facilitates computers to read and interpret previously existing data automatically. Machine learning can be used to classify diseases, determine the type of disease, and predict a disease. The machine learning method used is a decision tree. The step is to collect a data set consisting of input variables in the form of symptoms and output variables in the form of types of diarrheal diseases. Then change the form of the data into a table which becomes a model tree by looking for the entropy and gain of each symptom. After the model is formed, change the tree into a rule and simplify the rule (pruning). The results of identifying diarrheal diseases using the decision tree method were excellent, with an accuracy of 96.4%. Then a simplification of the rules is also obtained from 20 rules from 17 symptoms to 5 rules with a depth of 4 levels. The most influential symptoms in forming rules are G10, G14, and G16.