Title: The weighted Markov chain probability model: Forecasting discrete time sequence data
Authors: Nihan Potas - AHBV University (Turkey) [presenting]
Abstract: Statistical theory for the weighted Markov Chain Probability Model of presented. The aim is to discuss the theory and the application of these models to sets of time sequence data which will be summarized using the contingency table form. In the real data application, the grades point average for 8 semesters and cumulative-grades point average for 4 years of 1217 under-graduate students, beginning in the academic year 2013-2014, studying in Faculty of Political Science, Science and Engineering departments of Ankara University were used. Whether the change in students achievement status measurable in8 semesters and 4 years were evaluated. Markov chains can be an effective method for the description of the models, which will enhance capturing the use of forecast dynamic behavior with connection to the stochastic component. Such a forecasting analysis depends on the Markov Chain Theory, which is widely known, as it needs to defect correction that can be used to solve the inaccuracy and impracticability problems likely to arise from the forecast. With this perspective, the weight values for every state can be calculated using the Weighted Markov Chain Transition probability. The analysis clearly illustrates the benefits of weighted Markov Chain Probability model which is validated through accurate and reliable results obtained.