With the complexity of decision environments, the coexistence of heterogeneous decision information is more and more common. The paper explores the transformation method to unify heterogeneous information into a set of random values by segmenting the data range into several subintervals based on its distribution characteristics. This method can improve the consistency of distribution structure between the transformed values and initial information as much as possible. Through large-scale simulation, the solution method and conclusion of the optimal sub-segment interval in the random transformation process under different situations are summarized, which can provide references to the determination of the sub-segment sampling interval in practice. Further, the integrated aggregation method is fused with the random transformation process to obtain the superiority-based probability matrix, from which the ranking-based probability matrix and the possibility ranking result are deduced. This type of ranking result achieves the flexible transformation of decision conclusion from “absolute form” to “relative form”, which is more consistent with the diversity, uncertainty and fuzziness of the original decision information and is more interpretable and acceptable. Finally, the effectiveness of the proposed method is verified by comparison analysis. An application example, regarding comprehensive prediction of provincial development potential, is constructed to illustrate the characteristics and application prospect of this method. This study further enriches the transformation approach of heterogeneous information and provides method support for complex decision-making problems with multiple types of information co-existing.