Disease transmission models play a large role in determining appropriate containment strategies for many infectious disease outbreaks, including the recent COVID-19 epidemic. However, virus transmission often differs between cities. To account for this, a two-level spatial individual-level model (ILM) is used to capture the disease spread between and within cities. We then investigate the use of classification and regression ensemble methods using a random forest algorithm to estimate the two-level ILM parameters, rather than traditional yet computationally expensive procedures such as Bayesian MCMC. We compare the ability of these methods to estimate the parameters for the two-level ILM, and whether the resulting parameter estimates can successfully model the disease transmission.