A Markov Chain Analysis of Synthetic Film Genre Transitions on Streaming Platforms

Abstract

This study examines a first-order Markov chain as an interpretable model of transitions between film genres. The analysis uses a supplied synthetic frequency table containing 340 transitions among Action, Comedy, Horror, Romance, and Fantasy; it does not use observed streaming-platform records. Transition probabilities were obtained by normalizing each row of the count matrix, and the stationary distribution was calculated from the full-precision probabilities under a time-homogeneous model. The resulting long-run probabilities were 0.231016 for Action, 0.216088 for Comedy, 0.211556 for Fantasy, 0.175860 for Horror, and 0.165480 for Romance. Action had the largest stationary share and Romance the smallest. All states communicated, and the existence of return paths of lengths two and three established aperiodicity despite zero diagonal entries. The example therefore had a unique stationary distribution to which the modeled genre distribution converged. These results describe the supplied synthetic construction and do not establish actual audience preferences or recommendation accuracy. The model offers a transparent baseline for explaining genre transitions, but empirical use would require documented event sequences, validation of modeling assumptions, and comparison with temporal and personalized recommendation methods.