Prediction Of E-Payment Channels with Hidden Markov Model

Hidden Markov models (HMM) are a class of stochastic modelling that describe the evolution of observable events that depends on internal factors (hidden states) which are not directly observable. This probabilistic model has use in a wide range of disciplines, including engineering, finance, and medicine. In this study, HMM is employed in e-payment transactions to see the trends of these e-payment systems and to determine what would be the future trend of the e-payment channels. Two e-payment channels were considered: Automated teller machine (ATM) and point of sale (PoS) machines which are the observable state while the hidden states are high and a low volume of transactions. The emission matrix was determined based on the observable state count and the transition probability count. The filtering method was employed in the HMM to estimate the parameters of the model. Empirical analysis revealed that the volume of transactions for PoS surpass that of ATM based on the data used in this study. The steady-state probability also shows that in the future the volume of transactions for PoS with the probability of 0.875 (87.5%) would surpass that of ATM with the probability of 0.125 (12.5%).