Abstract:

Aim: The aim of this study is to examine and analyze new patterns of cesarean section anesthesia types and prediction performances of decision trees with data mining techniques. Materials and methods: Classification and clustering analysis were performed to analyze the data of 300 patients. Gini algorithm and C5.0 algorithm were applied to the data set with 24 parameters. These algorithms were also applied to the 16-parameter data set obtained after preprocessing. The estimation performances obtained were compared according to the accuracy criterion. Then, clustering analysis was applied to the 24 and 16 parameter data sets with the K-prototype algorithm. Results: The study revealed that the prediction success of the Gini algorithm was determined as 96.61%, and the prediction success of the pruned decision tree obtained by the Gini algorithm was 94.91%. The prediction success of the C5.0 algorithm was determined as 98.87%.In the clustering analysis performed with the K-prototype algorithm, the number of clusters was determined as 4 and 5 for both data sets, based on expert opinion, and important patterns were observed with these cluster numbers. Conclusion: As a result of the study, it was revealed that the C5.0 algorithm had the highest performance with an accuracy rate of 98.87As a result of the cluster analysis, it was concluded that the age of the patients, the duration of the operation, the type of previous anesthesia, the number of previous cesarean sections, the fear of anesthesia and the previous surgical operations were effective on the type of anesthesia in cesarean section cases.

Keywords: