Implementation of Decision Tree C4.5 for Big Five Personality Predictions with TF-RF and TF-CHI2 on Social Media Twitter

Every tweet can provide an information about someone's personality. The problem is how to classify an obtained information on Twitter social media into classes that will be created with good performance value. Personality classification through Twitter social media has been studied by several researchers. For instance, in Damanik Agnes and Khodra Masayu [3], Classification method using Support Vector Regression (SVR) and TF-IDF weighting method. The result shown good with Mean Absolute Error (MAE) 0.2739. In this paper, writer builds a system that classifies someone personality on Twitter social media using decision tree C4.5 classification method with TF-RF and TF-CHI₂ weighting method. The diffences on this paper is the weighting of each word using the weighting method TF-RF and TF-CHI₂ with the addition of new features for approaches based on user social behavior which will be explained in more detail in section II. From the results of the experiment on the 90% training data ratio and 10% test data (90:10) and the combination of personality features based on social behavior with a linguistic approach with TF-RF weighting obtained the accuracy results of 65.72%.