Optimalisasi algoritme voting feature intervals 5 menggunakan algoritme genetik pada data tuberkulosis paru

Abstract

We proposed a genetic algorithm to optimize VFI5 classification algorithm and to get the best feature weights in lung tuberculocis data. This research used a uniform value for this features weights which equals to one. The accuracy obtained was 83%. Genetic Algorithm (GA) is used to optimize VFI5 by determining the optimal weights for each feature. GA will combine each weight and search the best combination to get an optimal solution. In this research, GA can find the optimal weight feature VFI5 classification algorithm in lung tuberculocis data. The optimal feature weights are “the blood cough with average weight 0.91, limp with average weight 0.82, lost of appetite with average weight 0.63, body weight decrease with average weight 0.62, fever and perfiration with average weight 0.62”. in this reseached, 3-fold cross validation was used to divide data into traning and testing and obtained 95% accuracy in each fold. In conclucions, this research has provided with and accurate model to predict TB and Non TB data instances.