Application of Random Forest Method for Television Malfunction Prediction

Abstract

 In repairing a television (TV), it is necessary to understand the symptoms experienced by the TV. Therefore, technicians need to conduct an initial analysis of the causes of these symptoms. Analysis of the causes of TV damage can be predicted using a technological approach, one of which is by using an expert system. This study will focus on developing an expert system to predict the causes of TV damage. This study will apply the Random Forest method to predict TV damage based on historical datasets obtained from company X. Company X is a company engaged in the repair of electronic devices, one of which is TV. The data obtained will be used as training data to create a model that can predict the causes of TV damage. Then the experiment was carried out with a quantitative approach with experiments to optimise the model in increasing prediction accuracy. The model was evaluated using accuracy metrics. The results of the study showed that Random Forest has very good performance in classifying the causes of TV damage with a high level of accuracy reaching 100%. However, this study is only limited to certain historical data and does not consider external factors that influence damage to the TV.