Genetic Algorithm Optimization on Nave Bayes for Airline Customer Satisfaction Classification

Abstract

Airline companies need to provide satisfactory service quality so that people do not switch to using other airlines. The way that can be used to determine customer satisfaction is to use data mining techniques. Currently, the website www.kaggle.com has provided Airline Passenger Satisfaction data consisting of 22 attributes, 1 label and 25976 instances which are included in the supervised learning data category. Based on several previous studies, the Naïve Bayes algorithm can provide better classification performance than other classification algorithms. Several studies also state that the use of Naive Bayes can be optimized using Genetic Algorithm (GA) to obtain better performance. The use of Genetic Algorithm for Nave Bayes optimization in classifying Airline Passenger Satisfaction data requires further research to ensure the performance of the given classification. This study aims to compare the use of the Naive Bayes algorithm for the classification of Airline Passenger Satisfaction with and without GA optimization. The data validation process used in this study is to use split validation to divide the dataset into 95% training data and 5% testing data. The test results show that the use of GA on Naive Bayes can improve the classification performance of Airline Passenger Satisfaction data in terms of accuracy and recall with an accuracy value of 85.99% and a recall of 87.91%.

Keywords

• data mining
• classification
• naïve bayes
• genetic algorithm
• customer satisfaction