Traditional multiobjective optimization evolutionary algorithms (MOEAs) with a limited population have difficulty solving multiobjective optimization problems (MOPs), especially when the number of objectives is larger than three. In real-world applications, the decision maker (DM) may only prefer a partial region of the Pareto front (PF), named the region of interest (ROI). With the preference information from the DM, the complexity of solving MOPs can be greatly reduced since only solutions inside the ROI need to remain. This paper develops an approach of establishing preference model through the preference information —— the proportion of each objective. The approach can control the spread of the preferred solutions accommodating the DM's expectation of the extent of the ROI without the size of the ROI. Meanwhile, we incorporate the preference model into decomposition-based evolutionary multiobjective optimization methods, which means it is able to lead the search process toward the ROI. Extensive experiments show that our proposed method has effectiveness and competitiveness for approaching the preferred solutions in the ROI compared with four other preference-based algorithms.