Accurately determining geomechanical parameters and rock mass deformation are key components of slope stability assessment as well as structural design and safe construction. Numerical models are therefore commonly applied to these issues as a component of back analysis, which provides a practical tool for determining geomechanical parameters in rock engineering. Although once quantified, rock mass deformation can also be calculated using an appropriate numerical model, the traditional approaches tend to be time-consuming and cannot be used to accurately determine deformation uncertainty in practical rock engineering. A relevance vector machine (RVM) was adopted in this study to build the complex mapping relationship between geomechanical parameters and slope deformation, while an artificial bee colony (ABC) was also utilized to search for optimal parameters based on the RVM model and displacement monitored in the field. On this basis, both deformation and associated uncertainty can be predicted using this model via the input of determined geomechanical parameters. The method advocated in this paper was then applied to a rock slope at the Longtan hydropower station, China; results demonstrate that the RVM model accurately describes the relationship between geomechanical parameters and displacement increments and can thus be used to predict both deformation and associated uncertainty.