Browsing by Author "Koyuncu, Murat"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
Item A HYBRID METHOD FOR MISSING VALUE IMPUTATION(2022-02-16) Al-Brge, Basma; Koyuncu, MuratMissing data arises in almost all serious statistical analyses. Statistical analyses have a variety of methods to handle missing data, including some relatively simple approaches that can often yield reasonable results such as the random imputation approach. The missing data imputation process must be modeled in order to perform imputations correctly. Using datasets in empirical applications is very common to perform some tasks; however, missing values in datasets should be extracted from the datasets or should be estimated before they are used for processing to produce correct association rules or clustering in the preprocessing stage of data mining and processing. In this thesis, a hybrid approach is used that combines K-Nearest Neighbor (KNN) with Singular Value Decomposition (SVD) algorithm to improve the data imputation and produce data with high correlation with original missing values. The test results of the proposed hybrid method are compared with the results of several alternative methods for different rate of missing values and the results of the proposed method yields better performance than the others. The results are also compared with the reported results in the literature to give an idea about its performance.Item ESTIMATION ON OPTIMAL SIZE OF MICROSERVICES(2022-01-24) Vural, Hülya; Koyuncu, MuratCloud computing is becoming the de-facto standard for enterprises. The monolith applications of the past do not measure up to cloud standards. As a result, less coupled, more agile Microservices Architecture style has emerged. The Microservices Architecture proposes the use of smaller services when compared to traditional service oriented architectures. Since we were introduced to Microservices Architecture, there is an ongoing debate about what the actual size of a microservice should be. This thesis aims to identify what measures would help in finding the optimal size of a microservice. To achieve this aim, two different domain-driven design examples were taken for empirical analysis. Then those examples were modified to get more granular and less granular microservices. In the end, the more granular, original and less granular examples were compared. During the comparison, afferent coupling, efferent coupling, relational cohesion and COSMIC function points were calculated and compared. The results indicate that afferent and efferent coupling and relational cohesion values are better for deciding the optimal size. Based on the results, it is concluded that the domain-driven design can be used to obtain the optimal service granularity of microservices.