Notice: Undefined index: linkPowrot in C:\wwwroot\wwwroot\publikacje\publikacje.php on line 1275
[107660] Artykuł: Business Intelligence in Airline Passenger Satisfaction Study - a Fuzzy-Genetic Approach with Optimized Interpretability-Accuracy Trade-OffCzasopismo: Applied Sciences Tom: 11, Zeszyt: 11, Strony: 1-22ISSN: 2076-3417 Opublikowano: 2021 Autorzy / Redaktorzy / Twórcy Grupa MNiSW: Publikacja w czasopismach wymienionych w wykazie ministra MNiSzW (część A) Punkty MNiSW: 100 Pełny tekst DOI Keywords: business intelligence  airline passenger satisfaction  fuzzy rule-based systems  multiobjective evolutionary optimization  accuracy-interpretability trade-off optimization  |
The main objective and contribution of this paper is the application of our knowledge discovery business-intelligence technique (fuzzy rule-based classification systems) characterized by genetically optimized interpretability-accuracy trade-off (using multi-objective evolutionary optimization algorithms) to decision support related to airline passenger satisfaction problems. Recently published and accessible at Kaggle’s repository airline passengers satisfaction data set containing 259760 records is used in our experiments. A comparison of our approach with an alternative method (using SAS-system’s accuracy-oriented prediction tools to determine the attribute importance hierarchy) is also performed showing the advantages of our method in terms of: (i) discovering the actual hierarchy of attribute significance for passenger satisfaction and (ii) knowledge-discovery system’s interpretability-accuracy trade-off optimization. The main results and findings of our work include: (i) an introduction of the modern fuzzy-genetic business intelligence solution characterized both by high interpretability and high accuracy to the airline passenger satisfaction decision support, (ii) an analysis of the effect of possible "overlapping" of some input attributes over the other ones in order to discover the real hierarchy of influence of particular input attributes upon the airline passengers satisfaction, and (iii) an extended cross validation experiment confirming high effectiveness of our approach for different learning-test splits of the data set considered.