Endoscopic procedures for diagnosing gastrointestinal tract findings depend on specialist experience and inter-observer variability. This variability can cause minor lesions to be missed and prevent early diagnosis. In this study, deep learning-based hybrid stacking ensemble modeling has been proposed for detecting and classifying gastrointestinal system findings, aiming at early diagnosis with high accuracy and sensitive measurements and saving workload to help the specialist and objectivity in endoscopic diagnosis. In the first level of the proposed bi-level stacking ensemble approach, predictions are obtained by applying 5-fold cross-validation to three new CNN models. A machine learning classifier selected at the second level is trained according to the obtained predictions, and the final classification result is reached. The performances of the stacking models were compared with the performances of the deep learning models, and McNemar’s statistical test was applied to support the results. According to the experimental results, stacking ensemble models performed with a significant difference with 98.42% ACC and 98.19% MCC in the KvasirV2 dataset and 98.53% ACC and 98.39% MCC in the HyperKvasir dataset. This study is the first to offer a new learning-oriented approach that efficiently evaluates CNN features and provides objective and reliable results with statistical testing compared to state-of-the-art studies on the subject. The proposed approach improves the performance of deep learning models and outperforms the state-of-the-art studies in the literature.

Early diagnosis of diabetes, which can cause death, is very important for the health of the person. In the literature, machine learning techniques are frequently used in diagnosis of many diseases, including diabetes. The aim of the study is to predict diabetes with high accuracy by using machine learning and preprocessing techniques. Pima Indian Diabetes dataset was used in the study. J48 (Decision Tree), Naïve Bayes, Support Vector Machine, Logistic Regression, Multilayer Perceptron, K Nearest Neighbor, Logistic Model Tree, and Random Forest were used for classification. Of the preprocessing methods, feature selection, imputing missing values, normalization and standardization are performed. According to the results obtained, the highest accuracy value got with the Random Forest algorithm as 80.869.

Effort estimation of software testing plays a vital role in the effective completion of testing. In particular, software test effort estimation in defense projects is not an easy and simple phenomenon, owing to internal and external factors. This study aims to investigate the relationships between software test metrics used in the industry and software testing efforts. A method for estimating the testing effort is proposed using a novel set of software testing metrics that have not been used in any previously proposed software test effort estimation methods. In this study, 15 completed software projects of a CMMI Level-3 certified defense industry company were analyzed. The results of the empirical study show that the proposed method with the given metrics provides acceptable prediction quality, with Pred (0.25) and Pred (0.30) values equal to 0.867. We obtained plausible results and demonstrated that our newly proposed metrics can be safely used for software test effort estimation.

It is an important issue in the software industry to predict how much effort will be required for a software project as early as possible. Software size is one of the commonly used attributes in effort estimation. In this paper, we propose an early software size and effort estimation method based on conceptual model of the problem domain. Our method utilizes the noteworthy domain concepts identified mainly from the use cases written in the requirements phase of the software development lifecycle. In order to develop the model and evaluate its prediction quality, the use cases written and the effort data collected for 14 industrial software development projects of a CMMI level 3 certified defense industry company have been used. Evaluation results reveal a high correlation between the number of conceptual classes identified (i.e., domain objects) during the requirements analysis, the number of classes constituting the resulting software and the actual effort spent. Moreover, we have used the use case point (UCP) method to estimate the effort needed for each project and compared the results of UCP analysis with the results obtained with our method. The comparisons have shown that, for the projects considered, our method gives a better effort estimation compared to the effort estimated by using the UCP method.

The cohesion value is one of the important factors used to evaluate software maintainability. However, measuring the cohesion value is a relatively difficult issue when tracing the source code manually. Although there are many static code analysis tools, not every tool measures every metric. The user should apply different tools for different metrics. In this study, besides the use of these tools, we predicted the cohesion values (LCOM2, TCC, LCC, and LSCC) with machine learning techniques (KNN, REPTree, multi-layer perceptron, linear regression (LR), support vector machine, and random forest (RF)) to solve them alternatively. We created two datasets utilizing two different open-source software projects. According to the obtained results, for the LCOM2 and TCC metrics, the KNN algorithm provided the best results, and for LCC and LSCC metrics, the REPTree algorithm was the best. However, out of all the metrics, RF, REPTree, and KNN had close performances with each other, and therefore any of the RF, REPTree, and KNN techniques can be used for software cohesion metric prediction.