COST SENSITIVE LEARNING ALGORITHMS

This thesis studies the cost sensitive learning algorithms that calculate the class learning algorithms errors and costs. Data mining is the automated extraction of hidden predictive information from databases that can be applied to predict and diagnose many illnesses. Specifically, accurate classification of illnesses is a very important issue for the treatment of illnesses. The goal of classification is to build a set of models that can correctly predict the class of the different objects. Some algorithms produce better results than others. It is necessary to analyze systematically the performance of classifiers using a variety of datasets. In this thesis, many features were explored and 10 datasets were classified by using 5 classification algorithms. Logistic Discrimination Algorithm (LD), K-Nearest Neighbor Algorithm (KNN), Multilayer Perceptron Algorithm (MLP) and Nearest Mean algorithm and Decision Tree (C4.5) algorithms have been used for classification. These methods are applied to large and small datasets and then a large number of experiment results were obtained. The results show that there is no single algorithm that performs well in al