부산대학교 도서관

In the context of aviation, the anticipation of visibility is contingent upon the consideration of diverse meteorological factors. This study systematically examines the influence of the cross-validation technique ($k$) on the precision of visibility predictions, as gauged by root mean square error and mean absolute error. Employing the Regression Learner, encompassing 26 predetermined algorithms, and employing cross-validation ($k$) iterations ranging from 5 to 15, the primary objective was to discern the optimal model for visibility prognosis. Notably, our analysis extends to two distinct airports in Peninsular Malaysia, thereby enabling a comparative assessment. Results elucidate that the Gaussian Process Regression model consistently demonstrates superior efficacy across varied meteorological parameters and diverse $k$ values. The outcomes of this study are poised to yield practical implications, particularly in refining visibility prognostications and mitigating the likelihood of aviation incidents.