Kemal Akyol

Work place: Computer Engineering, Kastamonu University, Kastamonu, 37100, Turkey



Research Interests: Decision Support System, Data Mining


Kemal Akyol, He received his B.Sc. in Computer Science Department from Gazi University, Ankara/Turkey in 2002. He received his M.Sc. degree from Natural and Applied Sciences, Karabuk University, Karabuk/Turkey and Ph.D. degree from the same department. His research interests include data mining, decision support systems and expert systems.

Author Articles
Comparing the Performances of Ensemble-classifiers to Detect Eye State

By Kemal Akyol Abdulkadir Karaci

DOI:, Pub. Date: 8 Dec. 2022

Brain signals required for the brain-computer interface are obtained through the electroencephalography (EEG) method. EEG data is used in the analysis of many problems such as epileptic seizure detection, bipolar mood disorder, attention deficit, and detection of the sleep state of the vehicle driver. It is very important to determine whether the eye is open or closed, which is a substantial organ for the determination of the cognitive state of the person. The aim of this paper is to present a stable and successful model for detecting the eye states that are opened or closed. In this context, the performances of several ensemble classifiers were examined on the Emotiv EEG Neuroheadset dataset, which has 14 features excluding the target variable, 14980 records that have 8225 eye states opened and 6755 eye states closed. In the experiments, firstly the min-max normalization process was applied to the dataset, and then the classification performances of these classifiers were evaluated via a 5-fold cross-validation technique. The performance of each model was measured using accuracy, sensitivity, and specificity metrics. The obtained results show that the Random Forest algorithm is an acceptable level with 92.61% value of accuracy, 94.31% value of sensitivity and 91.36% value of specificity for detecting the eye state.

[...] Read more.
Care4Student: An Embedded Warning System for Preventing Abuse of Primary School Students

By Kemal Akyol Abdulkadir Karaci Muhammed Emin Tiftikci

DOI:, Pub. Date: 8 Aug. 2022

Child abuse is a social and medical problem that has negative effects on the individual development of the child and can lead to mental disorders such as depression and post-traumatic stress disorder in both short and long-term mental health. Therefore, any abuse that the child may encounter should be immediately intervened. This paper presents the design of an integrated embedded warning system that includes an embedded system module, a server-based module, and a mobile-based module as a solution to concerns of ensuring the safety of students in places where there are fewer safety measures. Our solution aims to ensure that the school management team is quickly informed about the adverse situation that primary school students may encounter and able to respond to them. In this context, this system activates the warning status when it correctly detects the phrases 'help me' and 'give it up'. Thus, any negativity that may be encountered in a closed environment is prevented. The embedded warning system detected correctly the phrase "help me" with 80%, and the phrase "give it up" with 75%.

[...] Read more.
Forecasting of Dry Freight Index Data by Using Machine Learning Algorithms

By Kemal Akyol

DOI:, Pub. Date: 8 Aug. 2019

Discovery of meaningful information from the data and design of an expert system are carried out within the frame of machine learning. Supervised learning is used commonly in practical machine learning. It includes basically two stages: a) the training data are sent to as input to the classifier algorithms, b) the performance of pre-learned algorithm is tested on the testing data. And so, knowledge discovery is carried out through the data. In this study, the analysis of Lloyd data is performed by utilizing Gradient Boosted Trees and Multi-Layer Perceptron learning algorithms. Lloyd data consist of the Baltic Dry Index, Capesize Index, Panamax Index and Supramax Index values, updated daily. Accurate prediction of these data is very important in order to eliminate the risks of commercial organization. Eight datasets from Lloyd data are obtained within the frame of two scenarios: a) the last three index values in the freight index datasets; b) the last three index values in both crude oil price and freight index datasets. The results show that the models designed with Gradient Boosted Trees and Multi-Layer Perceptron algorithms are successful for Lloyd data prediction and so proved their applicability.

[...] Read more.
A Study on Test Variable Selection and Balanced Data for Cervical Cancer Disease

By Kemal Akyol

DOI:, Pub. Date: 8 Sep. 2018

Cancer is a pestilent disease. One of the most important cancer kinds, cervical cancer is a malignant tumor which threats women's life. In this study, the importance of test variables for cervical cancer disease is investigated by utilizing Stability Selection method. Also, Random Under-Sampling and Random Over-Sampling methods are implemented on the dataset. In this context, the learning model is designed by using Random Forest algorithm. The experimental results show that Stability Selection, Random Over-Sampling and Random Forest based model are more successful, approximately 98% accuracy.

[...] Read more.
Diabetes Mellitus Data Classification by Cascading of Feature Selection Methods and Ensemble Learning Algorithms

By Kemal Akyol Baha sen

DOI:, Pub. Date: 8 Jun. 2018

Diabetes is a chronic disease related to the rise of levels of blood glucose. The disease that leads to serious damage to the heart, blood vessels, eyes, kidneys, and nerves is one of the reasons of death among the people in the world. There are two main types of diabetes: Type 1 and Type 2. The former is a chronic condition in which the pancreas produces little or no insulin by itself. The latter usually in adults, occurs when insulin level is insufficient. Classification of diabetes mellitus data which is one of the reasons of death among the people in the world is important. This study which successfully distinguishes diabetes or normal persons contains two major steps. In the first step, the feature selection or weighting methods are analyzed to find the most effective attributes for this disease. In the further step, the performances of AdaBoost, Gradient Boosted Trees and Random Forest ensemble learning algorithms are evaluated. According to experimental results, the prediction accuracy of the combination of Stability Selection method and AdaBoost learning algorithm is a little better than other algorithms with the classification accuracy by 73.88%.

[...] Read more.
A Study on Diagnosis of Parkinson’s Disease from Voice Dysphonias

By Kemal Akyol

DOI:, Pub. Date: 8 Jun. 2018

Parkinson disease that occurs at older ages is a neurological disorder that is one of the most painful, dangerous and non-curable diseases. One symptom that a person may have Parkinson’s disease is trouble that can occur in the voice of a person which is so-called dysphonia. In this study, an application based on assessing the importance of features was carried out by using multiple types of sound recordings dataset for diagnosis of Parkinson disease from voice disorders. The sub-datasets, which were obtained from these records and were divided into 70-30% training and testing data respectively, include the important features. According to the experimental results, the Random Forest and Logistic Regression algorithms were found successful in general. Besides, one or two of these algorithms were found to be more successful for each sound. For example, the Logistic Regression algorithm is more successful for the ‘a’ voice. The Artificial Neural Networks algorithm is more successful for the ‘o’ voice.

[...] Read more.
A Study on the Diagnosis of Parkinson’s Disease using Digitized Wacom Graphics Tablet Dataset

By Kemal Akyol

DOI:, Pub. Date: 8 Dec. 2017

Parkinson Disease is a neurological disorder, which is one of the most painful, dangerous and non-curable diseases, which occurs at older ages. The Static Spiral Test, Dynamic Spiral Test and Stability Test on Certain Point records were used in the application which was developed for the diagnosis of this disease. These datasets were divided into 80-20% training and testing data respectively within the framework of 10-fold cross validation technique. Training data as the input data were sent to the Random Forest, Logistic Regression and Artificial Neural Networks classifier algorithms. After this step, performances of these classifier algorithms were evaluated on testing data. Also, new data analysis was carried out. According to the results obtained, Artificial Neural Networks is more successful than Random Forest and Logistic Regression algorithms in analysis of new data.

[...] Read more.
A Study on Liver Disease Diagnosis based on Assessing the Importance of Attributes

By Kemal Akyol Yasemin Gultepe

DOI:, Pub. Date: 8 Nov. 2017

Liver is a needful body organ that forms an important barrier between the gastrointestinal blood, which contains large amounts of toxins, and antigens. Liver diseases contain hepatitis B and hepatitis C virus infections, alcoholic liver disease, nonalcoholic fatty liver disease and associated cirrhosis, liver failure and hepatocellular carcinoma are primary causes of death. The main purpose of this study is to investigate which attributes are important for effective diagnosis of liver disorders by performing the machine learning approach based on the combination of Stability Selection and Random Forest methods. In order to generate more accuracy, dataset was balanced by utilizing the Random Under-Sampling method. Important ones in all attributes were detected by utilizing the Stability Selection method which was performed on sub-datasets, which were obtained with 5 fold cross-validation technique. By sending these datasets to the Random Forest algorithm, the performance of the proposed approach was evaluated within the frame of accuracy and sensitive metrics. The experimental results clearly show that the Random Under-Sampling method can potentially improve the performance of the combination of Stability Selection and Random Forest methods in machine learning. And, the combination of these methods provides new perspectives for the diagnosis of this disease and other medical diseases.

[...] Read more.
Assessing the Importance of Attributes for Diagnosis of Diabetes Disease

By Kemal Akyol

DOI:, Pub. Date: 8 Sep. 2017

Diabetes is a chronic, metabolic disease related to the rise of levels of blood glucose. According to the current data from the World Health Organization, 422 million adults have diabetes in the world and prevalence of diabetes is 13.2%. Disregarding the diagnosis and treatment of the disease leads to some major problems on kidneys, heart and blood vessels, eyes, nerves, pregnancy and wound healing. The most common type of diabetes and usually in adults, Type 2 diabetes occurs when the body becomes resistant to insulin or does not make enough insulin. The main objective of this study is to make more successful this disease by investigating the important attributes based on assessing the importance of attributes using the Stability Selection method. The proposed method might be a powerful tool for the importance of attributes, and effective diagnosis of this disease with the classification accuracy is 78.57% and ROC value is 0.75.

[...] Read more.
Other Articles