Akademik Veri Yönetim Sistemi
Türkiye Tıbbi Bilişim Kongresi Derneği, Trabzon, Türkiye, 30 - 31 Mayıs 2024, ss.186-190
Problems that are hard and time-consuming to tackle in daily life can be solved with the help of optimization methods. In situations where traditional optimization methods are insufficient, heuristic and metaheuristic algorithms have been developed to identify solutions that are closest to the global answer. These algorithms can therefore be applied to parameter estimation as well. Appropriate parameter estimation is crucial when using logistic regression for classification. If a parameter is chosen incorrectly, the classification analysis may perform poorly since it will produce inaccurate predictions.In this study, the aim was to evaluate the combined use of logistic regression with metaheuristic algorithms. For this purpose, the parameters of the logistic regression model were estimated using metaheuristic algorithms, and the performance metrics of each model were assessed. The metaheuristic algorithms used for parameter estimation were PSO, ABC, and ALSO. The application data was obtained from Kaggle and split into training and test datasets. Five performance indicators were used to compare the analysis results: F-score, accuracy, precision, selectivity, and sensitivity.The research revealed that the system's logistic regression had the best accuracy value, which was 0.83. By contrast, the ALSO method yielded the highest accuracy value of 0.82 and the best F-score of 0.89.