Akademik Veri Yönetim Sistemi
Veri Bilimi Dergisi, cilt.4, sa.2, ss.57-61, 2021 (Hakemli Dergi)
At present, the number of passengers preferring to use the airline is increasing with eachpassing day. Thus, correctly analysing the airfare prices is essential to raise awareness ofpassengers. Some researchers have applied different kinds of Machine Learning (ML)algorithms to predict the airfare prices. However, to the best of our knowledge, penalizedregression methods have not been used to analyse the airfare prices. Ridge, Lasso, and ElasticNet regressions are penalized regression methods. The dataset used in this study consists of1814 one-way flights from Greece to Germany. The developed Ridge, Lasso, and Elastic Netmethods were achieved to provide convincing results for airfare prices analysis based on MeanSquared Error-MSE values (Ridge:160103, Lasso:159280, Elastic Net:174203) and MeanAbsolute Error-MAE values (Ridge:147.74, Lasso:146.43, Elastic Net:346.86). MSE and MAEexplain how close a regression line is to a set of points. They take the distances from the pointsto the regression line (refers to “errors”). MSE takes the squares of the errors, MAE takes theabsolutes of the errors. The lower they are, the better the prediction is. Thus, in our case, Lassoregression can be considered better than the ridge and elastic net due to the lowest MSE andMAE values. In other words, the results and findings reveal that the proposed Lasso method ispotentially better than the others in the analysis of datasets consisting of one-way flights.