Affected by the microaggregation of supercritical fluids, supercritical pyrolysis of hydrocarbon fuels shows unique product distributions and pathways. In this work, the role of inhomogeneous microaggregation during supercritical pyrolysis are clarified via computational method, with the benchmark of methylcyclohexane. Characteristic pyrolysis behaviors under supercritical conditions are intimately revealed and characterized through collective variable-driven hyperdynamics (CVHD) simulations. Liquid-like atoms, which representing the inhomogeneity of supercritical fluids are described by the trained machine-learning classifier, then proposed as the specific reason for the increase in the variety and frequency of bimolecular reactions. In addition, the intrinsic pyrolysis mechanism has also been confirmed to regulate the inhomogeneity effect on reactions through comparisons between different hydrocarbons.