Nowadays, there is a growing expectation to promote operational efficiency of ship fleets in maritime transportation. This study presents a methodical approach basis on statistical learning to model ship performance monitoring problem. It takes the advantage of shrinkage models such as Ridge and Lasso. The demonstrations are conducted through numerical operational features (i.e., fuel consumption, speed, trim, draft, heeling, headwind, etc.) via train and test data set recorded from 2-h voyages of a ferry ship in 2-month period. The findings address to consider trim, pitch, and wind effect. Besides feature selections, both models, which are capable of predicting overall accuracy, learn the relationship between features from the original data set. Comparing to current nonlinear models applied to ship operational performance problems, the Ridge and the Lasso models enable to minimize complexity and to enhance interpretability. Consequently, this study is capable of increasing situational awareness of ship operators (Master and Chief Engineer) and shore-based organizations for monitoring of ship performance. A promising future research may conduct interface designs to transform the continuously monitored features into control actions in terms of ship operational performance management concept.