Bünyamin Sarıbacak

Iron deficiency anemia is one of the most common types of anemia worldwide. In recent years, new developments in the field of medicine have offered early diagnosis and treatment opportunities for anemia patients. In the field of data science, in parallel with the developments in medicine, significant developments are taking place in subjects such as data collection, storage, processing, and reporting. Interdisciplinary joint studies positively contribute to patients’ quality of life and lifespan. In this study, the accuracy of the statistical results was tested with Machine Learning Method (MLM) while investigating the factors that affect the correct prediction of Iron Deficiency Anemia (IDA) diagnosis. In the first stage, the relationships between all variables in the data set and their effects on the differentiation of disease groups were investigated using univariate and multivariate statistical methods. In the second step, the data set was analyzed in detail using four different methods with Artificial Neural Network (ANN) classifier. Weka 3.8 application was preferred for these operations. In the last stage, the results obtained in both stages were compared. Accordingly, it has been observed that hemoglobin (Hb), mean cell volume (MCV), iron (Fe), and ferritin (FERR) have more effects on IDA. ANN (98.06%) is a better discriminator with a correct classification rate

Diagnosis of a disease is one of the most important processes in the field of medicine. Thus, computer-aided detection systems are becoming increasingly important to assist physicians. The iron deficiency anemia (IDA) is a serious health problem that requires careful diagnosis. Diagnosis of IDA is a classification problem, and there are various studies conducted. Researchers also use feature selection approaches to detect significant variables. Studies so far investigate different classification problems such as outliers, class imbalance, presence of noise, and multicollinearity. However, datasets are usually affected by more than one of these problems. In this study, we aimed to create multiple systems that can separate diseased and healthy individuals and detect the variables that have a significant effect on these diseases considering influential classification problems. For this, we prepared different datasets based on the original dataset whose outliers were removed using different outlier detection methods. Then, a multistep classification algorithm was proposed for each dataset to see the results under irregular and regulated conditions. In each step, a different classification problem is handled. The results showed that it is important to consider each question together as it can and should change the outcome. Dataset and R codes used in the study are available as supplementary files online.

Risk estimation is of great importance in financial risk management. In this study, the risk estimation of the exchange rate portfolio is performed via the stochastic copula approach. This model‐based latent process has a parameter that changes over time and thus can model the dependency structure between variables in a comprehensive and dynamic way. First, the marginals of the returns are handled with ARMA‐GARCH‐type models. Then, the dependency between variables is modeled via the stochastic copula approach. Finally, risk estimates are carried out at 95% and 99% confidence level for the foreign exchange portfolios. It is found that the proposed risk estimation model based on the stochastic copula approach outperforms both classical methods and static copula models.