A Technique for Efficient Estimation of Dynamic Structural Equation Models: A Case Study

Compartmental modeling for pandemic data analysis: The gap between statistics and models

“…Assuming that the current official data situation is hardly actually to change, the future work should involve the extension of the proposed method to models that offer increased modeling flexibility (e.g., models with stochastic latent part, as this would allow for discernment of the different types of noise present with data, or models with finer-grained compartments and relations between them). Such extensions would be characterized by more complex mathematics that could be handled in the framework of estimator recursioning technique [28] . Moreover, another promising future opportunity pertains to applying the proposed method to modeling based on overall excess mortality (instead of COVID-19 deaths) because factual deaths comprise an objective criterion and expected deaths are a rather well-implied one [44] .…”

“…The model defined by Equations (6) – (9) can be considered the ultimate mathematical object of the proposed method. The following (and final) step consists of the corresponding parameter estimation, for which we employ selected aspects of the estimator recursioning technique [28] and a standard optimization algorithm, thus effectively carrying out the maximum likelihood estimation. We consider the exact mathematical details to be of secondary importance for the message the paper is meant to deliver; hence, we present them only as supplementary material in the form of an application, which can also be used to reproduce results.…”

“…It is true that the maximum likelihood approach often proves to be hardly tractable (especially given high model complexity and a large amount of data), leading researchers to resort to various heuristics that are usually easier to compute but harm the quality of the estimates [27] . Thus, it must be explicitly stressed that efficient parameter estimation inevitably requires a certain level of effort to establish suitable mathematical apparatus [28] .…”