Conducting experiments to understand and model a complex process or system is usually costly and time-consuming due to multistages, multivariables, and multidisciplinary issues involved in the complex process. To reduce the complexity, for a single experiment, experimenters often fix some variables and investigate the effects of a smaller subset of variables. If then, it is possible to build individual models for each subset of variables, but this only allows partial understanding of the whole process. In this paper, we propose a method for building a holistic model of a complex process using multiple partial models that are learned from multiple sub-experiments that focus on different variables or the same variables but with different variable ranges. Using the proposed holistic model, it should be possible to provide an initial understanding of the complex process involving all variables. The effectiveness of the proposed method is demonstrated using a real example from a buckypaper process.