Mainstream machine learning approaches to predictive analytics consistently prove their ability to perform well using a variety of datasets, although the task of identifying an optimally-performing machine learning approach for any given dataset becomes much less intuitive. Methods such as ensemble and transformation modeling have been developed to improve upon individual base learners and datasets with large degrees of variance. Despite the increased generalizability and flexibility of ensemble approaches, the cost often involves sacrificing inference for predictive ability. This paper introduces an alternative approach to ensemble modeling, combining the predictive ability of an ensemble framework with localized model construction through the incorporation of cluster analysis as a pre-processing technique. The workflow not only outperforms independent base learners and comparative ensemble methods, but also preserves local inferential capability by manipulating cluster parameters and maintaining interpretable relative importance values and non-transformed coefficients for the overall consideration of variable importance. This paper demonstrates the ensemble technique on a dataset to estimate rates of health insurance coverage across the state of Missouri, where the cluster pre-processing assists in understanding both local and global variable importance and interactions when predicting high concentration areas of low health insurance coverage based on demographic, socioeconomic, and geospatial variables.

The purpose of this study was to evaluate the relationship between violent events targeted against aid workers and the incidence of Ebola virus disease (EVD) during the 2018-19 Democratic Republic of the Congo (DRC) outbreak in the North Kivu and Ituri provinces. Time series models using vector autoregression were constructed using violent event data confined to the outbreak region from the Armed Conflict Location and Event Dataset (ACLED), in combination with EVD incidence reporting from the World Health Organization and DRC Ministry of Health, to examine intervariable temporal relationships, paired with Granger causality testing to assess both uni- and multidirectional associations. Violent events against aid workers and Granger-causing EVD incidence were found to be significant across 2 principal lag-ranges, 8 to 14 days and 22 to 29 days, both suggesting plausible causal associations. The multivariate model for violent events and violence-related fatalities Granger-causing EVD incidence was also found to be significant at lags greater than 9, reinforcing the plausible causal association. Findings from the study suggested that the relationship between targeted violence and EVD incidence may be explained etiologically, as significant lag-ranges corresponded to plausible patient presentation timeframes, based on the incubation period for EVD. Additionally, the findings may also be explained through impact on operations, when events targeting facilities, supply lines, and personnel affect treatment capability.