Comparison Study of Logistic Regression Model for Albanian Texts

Abstract

Considering authorship attribution as a classification problem we attempt to estimate the probability to find the right author for each text under study. In this paper using R we first improve the simple model for six Albanian texts, (I) increasing number of texts and number of independent variables and then compare the results taken with them of the multinomial logistic regression (II). The model was applied on a set of one hundred texts of ten different authors. For all the authors under study the average correct predicted probability is 0.918. Analyzing data from different Albanian texts, results that about 40% of their letters consist of vowels. As conclusion comparing results taken with them of (II) multinomial logistic regression model for Albanian texts has more advantages than logistic regression model.