Applying Markov-based forecasting in enrolment planning

The education sector is a multidimensional and complex system, affected by numerous internal and external factors. Institutional planning in such a speculative environment demands appropriate tools, especially when forecasting and modeling the future is necessary. Predictive analytics can help executives to identify the likelihood of future outcomes of their institutions based on past and current data, as well as to consider internal and external influencing factors. Such analyses can utilize a number of approaches varying from simple statistical techniques, data mining and predictive modeling tools to advanced machine learning algorithms. Selecting an appropriate yet effective model for two samples of enrolment planning, is the goal of the current paper. The Markov Chain is a well-known technique to forecast stochastic time-series data and is used in the current research. The suggested model is a homogenous Markov Chain which is applied on modeling Course-enrolment. Generating the Transitional probability matrix is the core concept of the model. To achieve this, analyzing the historical data to identify all possible valid transitional states is the first essential phase. Calculating transitional probabilities among all states is the second major phase. We have utilized a frequentist approach to achieve the transitional probabilities. The rest is about computing the likelihood of possible future states by implementing different scenarios by way of tweaking the elements of the primary Transitional probability matrix and analyzing the results. In addition to its ability to forecast stochastic processes, another advantage of a homogenous Markov model is its simplicity in implementation.