Citation Number
16
Views
13
Downloands
3
Makine öğrenmesi biyoteknoloji alanından eğitim bilimlerine, doğal dil işlemeden duygu analizine kadar medikal, eğitim, işletme gibi birçok alanda aktif olarak kullanılan bir disiplindir. Kullanım alanı genişledikçe regresyon, sınıflama, kümeleme gibi farklı problemlere çözüm arayan makine öğrenmesi, iflas tahmini probleminde de kullanılmaya başlamıştır. Makine öğrenmesi disiplininde algoritma sayısı arttıkça, parametreler değiştikçe farklı doğruluk oranlarına ulaşmak mümkündür. Bu amaçla, çalışmada k En Yakın Komşu algoritmasına yer verilmiş farklı uzaklık ölçütleri (Euclidean, Manhattan, Chebysev, Minkowski) kullanılarak yapılan sınıflandırma işlemi sonucunda en yüksek doğruluk oranına sahip uzaklık ölçütü belirlenmiştir. Veri seti %70 eğitim- %30 test seti olarak bölünmüş çeşitli performans ölçütleri kullanılarak algoritmalar birbiriyle karşılaştırılmıştır.
Machine learning is a discipline that is actively used in many areas, from the field of biotechnology to educational sciences, from natural language processing to emotional analysis, such as medical, education, business. With the extension of the scope of use, machine learning that is looking for solutions to different problems such as regression, classification, accumulation, has also begun to be used in the problem of bankruptcy forecast. As the number of algorithms in the machine learning discipline increases, it is possible to
Machine learning is a discipline that is actively used in many areas such as medical, education and business management, from biotechnology to educational science, natural language processing to emotion analysis. As the area of use expanded, machine learning, which was looking for solutions to different problems such as regression, classing and clustering, also started to be used in the problem of bankruptcy prediction. As the number of algorithms increases in machine learning discipline, it is possible to achieve different accuracy rates as parameters change. For this purpose, the k Nearest Neighbor algorithm was involved in our study and the distance measure with the best accuracy were determined as a result of the classification process using different distance measures (Euclidean, Manhattan, Chebysev, Minkowski). The data set is divided into 70% training - 30% test; algorithms are compared using various performance criteria.
Journal Type : Ulusal
Article Detail 
Citation Owners
Similar Articles
Download PDF
Journal Information
Listen Paper
Print This Page
Share
Mendeley
