Recently, there has been a shift from modeling driven approach to data driven approach in Agent Based Modeling and Simulation (ABMS). This trend towards the use of data-driven approaches in simulation aims at using more and more data available from the observation systems into simulation models [1, 2]. In a data driven approach, the empirical data collected from the target system are used not only for the design of the simulation models but also in initialization, evaluation of the output of the simulation platform. That raises the question how to manage empirical data, simulation data and compare those data in such agent-based simulation platform. In this paper, we first introduce a logical framework for data driven approach in agent-based modeling and simulation. The introduced framework is based on the combination of Business Intelligence solution and a multi-agent based platform called CFBM (Combination Framework of Business intelligence and Multi-agent based platform). Secondly, we demonstrate the application of CFBM for data driven approach via the development of a Brown Plant Hopper Surveillance Models (BSMs), where CFBM is used not only to manage and integrate the whole empirical data collected from the target system and the data produced by the simulation model, but also to initialize and validate the models. The successful development of the CFBM consists not only in remedying the limitation of agent-based modeling and simulation with regard to data management but also in dealing with the development of complex simulation systems with large amount of input and output data supporting a data driven approach.