Support Vector Machine

Anders Hedegaard, Sebastian Kragh & Isabella Marlene Gerda Thuesen

Student thesis: Master thesis

Abstract

Support Vector Machine is a popular method used for classification. In this paper, we develop an understanding of the theory behind the method of Support Vector Machine and its application. We find, that Support Vector Machine uses functions called kernels, to map data into a higher dimensional space in order to separate data with a hyperplane using linear classification theory. In order to construct a classification model, that can be applied well to other similar data, data is split into a training set, on which the model is built and a test set, on which the accuracy of the model is evaluated. Due to the properties of kernels, and the limited number of observations used for the construction of the model, Support Vector Machine is an efficient and robust method. To illustrate its application, we apply the Support Vector Machine classification to a set of financial data from the Orbis database, where we develop a model for classifying bankruptcy of Hotels. We compare the accuracy of different kernels and investigate the computational power required by Support Vector Machine.

EducationsMSc in Mathematics , (Graduate Programme) Final Thesis
LanguageDanish
Publication date2018
Number of pages121
SupervisorsSøren Feodor Nielsen