An Effective Classification Method for Facebook Data
Fatih Ertam · International Journal of Applied Mathematics Electronics and Computers · 2017
Today, the use of the internet has become very common.One of the most important reasons for its widespread use is social media tools.Especially Facebook has a very important place in social media tools.For this study, classification was done by using Facebook data.Classifications made by artificial learning algorithms on a previously used data set are compared with accuracy values and learning times.For this purpose, support vector machines (SVM), extreme learning machines (ELM) and K nearest neighbor (kNN) approaches are compared.For the study, SVM and ELM algorithms were observed using different activation functions.For the study with KNN, different K values were tested with different distance metric calculation methods.In the classification approach with ELM, it was observed that higher accuracy values were reached in a shorter time.In addition, Receiver Operating Characteristic (ROC) curves are plotted for the classification in which the best values are obtained for each algorithm.