Yazar "Kilic, Umit" seçeneğine göre listele

Listeleniyor 1 - 5 / 5
  • [ X ]
    Öğe
    Artificial Bee Colony Algorithm for Feature Selection on SCADI Dataset
    (IEEE, 2018) Keles, Mumine Kaya; Kilic, Umit
    Data consists of relevant and irrelevant features. These irrelevant features may mislead classification or clustering algorithms. Feature selection algorithms are used to avoid that misleading and to obtain better results using fewer number of features than dataset has. The purpose of feature selection is to choose as small number of relevant features as possible to enhance the performance of the classification. In this paper, Artificial Bee Colony algorithm (ABC), which is proposed by karaboaa in 2005, is implemented as feature selection algorithm (ABC-FS) and used on Self-Care Activities Dataset based on ICF-CY (SCAM). SCADI that contains 206 attributes of 70 children with physical and motor disability is a dataset for self-care problems. Feature selection operation is carried out using Gain Ratio, Info Gain, and Chi Square. These selected features are utilized to obtain classification results. Then, features are selected by ABC-FS and results are compared. Seven of 206 features are selected by ABC-FS, and 88.5714% accuracy rate and 0.871 F-Measure value are obtained while best of Info Gain, Gain Ratio and Chi-Square is 84.2857 (1/0 and 0.824, respectively. The experimental results show that the selected features by ABC-FS generally have higher accuracy than the raw dataset and Info Gain, Gain Ratio and Chi-Square.
  • [ X ]
    Öğe
    Binary Anarchic Society Optimization for Feature Selection
    (Editura Acad Romane, 2023) Kilic, Umit; Sarac Essiz, Esra; Kaya Keles, Mumine
    Datasets comprise a collection of features; however, not all of these features may be necessary. Feature selection is the process of identifying the most relevant features while eliminating redundant or irrelevant ones. To be effective, feature selection should improve classification performance while reducing the number of features. Existing algorithms can be adapted and modified into feature selectors. In this study, we introduce the implementation of the Anarchic Society Optimization algorithm, a human-inspired algorithm, as a feature selector. This is the first study that utilizes the binary version of the algorithm for feature selection. The proposed Binary Anarchic Society Algorithm is evaluated on nine datasets and compared to three known algorithms: Binary Genetic Algorithm, Binary Particle Swarm Optimization, and Binary Gray Wolf Optimization. Additionally, four traditional feature selection techniques (Info Gain, Gain Ratio, Chi-square, and ReliefF) are incorporated for performance comparison. Our experiments highlight the competitive nature of the proposed method, suggesting its potential as a valuable addition to existing feature selection techniques.
  • [ X ]
    Öğe
    Feature Selection with Artificial Bee Colony Algorithm on Z-Alizadeh Sani Dataset
    (IEEE, 2018) Kilic, Umit; Kaya Keles, Mumine
    Relevant and irrelevant features compose data. Evaluation of these features is the fundamental task for classification and clustering processes and during this processes, irrelevant features induce obtaining false results. Likewise, due to the relevant features' direct effect on the processes, results can be more correct and stable. This also represents the aim of the feature selection process that tries to achieve as high as possible results with as small as possible feature selection subset. In this study, Artificial Bee Colony (ABC) algorithm based feature selection method is updated and employed on Z-Alizadeh Sani data set that consists of 56 features including the class attribute collected from 303 patients. 16 of the 56 features are selected by ABC based updated feature selection method. Also, accuracy and F-measure values are measured as 89.4% and 0.894 respectively, which are higher than the values produced by the raw dataset.
  • [ X ]
    Öğe
    PREDICTION OF CONCRETE STRENGTH WITH DATA MINING METHODS USING ARTIFICIAL BEE COLONY AS FEATURE SELECTOR
    (IEEE, 2018) Kaya Keles, Mumine; Keles, Abdullah Emre; Kilic, Umit
    Concrete which is a highly complex material is the most basic input of the construction industry. Because of its strength, concrete is one of the most preferred structural building materials. In the ready-mixed concrete sector, there is an increasing need for earthquake resistant structures due to the fact that some producers produce out of control and poor quality. Ready-mixed concrete is a product whose quality can only be understood at the end of the 28th day if it is only controlled by taking the sample by the user. In this study, a data mining study was conducted on the factors affecting the 28-day compressive strength of concrete using the Concrete Slump Test Data Set from UCI Machine Learning Repository. The Artificial Bee Colony Algorithm is used as a feature selection method in order to determine the important ones of the concrete components, which are cement, slag, fly ash, water, superplasticizer, coarse aggregate, and fine aggregate, affecting concrete strength and tried to predict the strength with data mining algorithms. As a result of the study, it was observed that Random Forest Algorithm gave the highest success rate with 91.2621% accuracy using only 3 features, which are cement, fly ash, and water. This means that it is possible to predict the compressive strength of concrete with a ratio above 90% by using a smaller number of concrete components.
  • [ X ]
    Öğe
    Proposed Artificial Bee Colony Algorithm as Feature Selector to Predict the Leadership Perception of Site Managers
    (Oxford Univ Press, 2021) Keles, Mumine Kaya; Kilic, Umit; Keles, Abdullah Emre
    Datasets have relevant and irrelevant features whose evaluations are fundamental for classification or clustering processes. The effects of these relevant features make classification accuracy more accurate and stable. At this point, optimization methods are used for feature selection process. This process is a feature reduction process finding the most relevant feature subset without decrement of the accuracy rate obtained by original feature sets. Varied nature inspiration-based optimization algorithms have been proposed as feature selector. The density of data in construction projects and the inability of extracting these data cause various losses in field studies. In this respect, the behaviors of leaders are important in the selection and efficient use of these data. The objective of this study is implementing Artificial Bee Colony (ABC) algorithm as a feature selection method to predict the leadership perception of the construction employees. When Random Forest, Sequential Minimal Optimization and K-Nearest Neighborhood (KNN) are used as classifier, 84.1584% as highest accuracy result and 0.805 as highest F-Measure result were obtained by using KNN and Random Forest classifier with proposed ABC Algorithm as feature selector. The results show that a nature inspiration-based optimization algorithm like ABC algorithm as feature selector is satisfactory in prediction of the Construction Employee's Leadership Perception.