“…Further, Shang (2016) uses marginal likelihood as a means of selecting the optimal semi-metric. Building on the early work by , and Shang (2013Shang ( , 2014aShang ( ,b, 2016Shang ( , 2020, we consider a kernel error-density estimator which explores data-driven features, such as asymmetry, skewness, and multi-modality, and relies on residuals obtained from the estimated regression function and bandwidth of residuals. Differing from those early work, we derive an approximate likelihood and a posterior for the functional partial linear model (a semiparametric model).…”