Yazar "Yanik, Seda" seçeneğine göre listele

Listeleniyor 1 - 4 / 4
  • [ X ]
    Öğe
    A neural networks approach to predict call center calls of an internet service provider
    (IOS Press BV, 2022) Namli, Özge H.; Yanik, Seda; Nouri, Faranak; Serap Şengör, N.; Koyuncu, Yusuf Mertkan; Uçar, Ömer Berk
    In today's competitive business environment, companies are striving to reduce costs and workload of call centers while improving customer satisfaction. In this study, a framework is presented that predicts and encourages taking proactive actions to solve customer problems before they lead to a call to the call center. Machine learning techniques are implemented and models are trained with a dataset which is collected from an internet service provider's systems in order to detect internet connection problems of the customers proactively. Firstly, two classification techniques which are multi perceptron neural networks and radial basis neural networks are applied as supervised techniques to classify whether the internet connection of customers is problematic or not. Then, by using unsupervised techniques, namely Kohonnen's neural networks and Adaptive Resonance Theory neural networks, the same data set is clustered and the clusters are used for the customer problem prediction. The methods are then integrated with an ensemble technique bagging. Each method is implemented with bagging in order to obtain improvement on the estimation error and variation of the accuracy. Finally, the results of the methods applied for classification and clustering with and without bagging are evaluated with performance measures such as recall, accuracy and Davies-Bouldin index, respectively. © 2022 - IOS Press. All rights reserved.
  • [ X ]
    Öğe
    A novel ensemble artificial intelligence approach for coronary artery disease prediction
    (Emerald Group Publishing Ltd, 2024) Namli, Ozge H.; Yanik, Seda; Erdogan, Aslan; Schmeink, Anke
    PurposeCoronary artery disease is one of the most common cardiovascular disorders in the world, and it can be deadly. Traditional diagnostic approaches are based on angiography, which is an interventional procedure having side effects such as contrast nephropathy or radio exposure as well as significant expenses. The purpose of this paper is to propose a novel artificial intelligence (AI) approach for the diagnosis of coronary artery disease as an effective alternative to traditional diagnostic methods.Design/methodology/approachIn this study, a novel ensemble AI approach based on optimization and classification is proposed. The proposed ensemble structure consists of three stages: feature selection, classification and combining. In the first stage, important features for each classification method are identified using the binary particle swarm optimization algorithm (BPSO). In the second stage, individual classification methods are used. In the final stage, the prediction results obtained from the individual methods are combined in an optimized way using the particle swarm optimization (PSO) algorithm to achieve better predictions.FindingsThe proposed method has been tested using an up-to-date real dataset collected at Basaksehir & Ccedil;am and Sakura City Hospital. The data of disease prediction are unbalanced. Hence, the proposed ensemble approach improves majorly the F-measure and ROC area which are more prominent measures in case of unbalanced classification. The comparison shows that the proposed approach improves the F-measure and ROC area results of the individual classification methods around 14.5% in average and diagnoses with an accuracy rate of 96%.Originality/valueThis study presents a low-cost and low-risk AI-based approach for diagnosing heart disease compared to traditional diagnostic methods. Most of the existing research studies focus on base classification methods. In this study, we mainly investigate an effective ensemble method that uses optimization approaches for feature selection and combining stages for the medical diagnostic domain. Furthermore, the approaches in the literature are commonly tested on open-access dataset in heart disease diagnoses, whereas we apply our approach on a real and up-to-date dataset.
  • [ X ]
    Öğe
    Improving Disease Diagnosis with Integrated Machine Learning Techniques
    (Springer International Publishing Ag, 2022) Namli, Ozge H.; Yanik, Seda
    As the digital transformation is constantly affecting every aspect of our lives, it is important to enhance and use machine learning models more effectively also in the healthcare domain. In this study, we focus on the application of machine learning algorithms for disease diagnosis in order to support decision making of physicians. Different classification methods are used to predict the diameter narrowing in the heart using an anonymous dataset. In order to increase the prediction ability of the machine learning algorithms, we employ different feature extraction methods such as Autoencoder, Stacked Autoencoder, Convolutional Neural Network, and Principal Component Analysis methods and integrate each feature extraction method with the classification methods. Then, we compare the prediction performances of individual and feature-extraction-integrated classification methods. It is shown that the prediction performance of the classification methods increase when integrated with feature extraction methods. However, it is concluded that not all feature extraction methods work as well with all classification methods. When a specific classification method is integrated with the appropriate feature extraction method, a better improvement in the prediction performance can be obtained.
  • [ X ]
    Öğe
    Improving Heart Disease Diagnosis: An Ensemble Machine Learning Approach
    (Springer International Publishing Ag, 2024) Namli, Ozge H.; Yanik, Seda
    Improving the performance of machine learning approaches in the field of health is of the utmost importance because early and correct diagnosis and treatment of diseases are essential for human life. From this point of view, an ensemble machine learning approach has been proposed for the diagnosis of heart disease within the scope of this study. In the first step of the proposed approach, feature extraction is performed using the Convolutional Neural Network on the dataset. In the next step, prediction results are obtained using individual classification methods such as Multi-layer Perceptron, Support Vector Machine, and Random Forest. Finally, the obtained prediction results are combined using the majority voting method. The results which are compared according to the critical classification performance criteria show that the proposed ensemble method gives better results than the individual methods. Heart disease can be predicted with an accuracy of 86.4% with the proposed ensemble approach.