In a mobile agent system, if agents’ functionality can be assessed and evaluated between peers of environmental modelling, it can reduce the exploration burden of unvisited states and unseen situations, thus an effectual learning process has to be accelerated. So as to construct an accurate and effectual model in certain time period is a significant problem, specifically in complex environment. To overcome this crisis, the investigation anticipates a model based data mining approach based on tree structure to achieve co-ordination amongst the mobile agent, effectual modelling and less memory utilization. The anticipated model suggests Mobi-X architecture to mobile agent system with a tree structure for effectual modelling. The construction of tree for real time mobile agent system is utilized to generate virtual experiences like elapse time during mining of tree structure. In addition, this model is appropriate for knowledge mining. This work is inspired by knowledge mining concept in mobile agent systems where an agent can built a global model from scattered local model held by individual agents. Subsequently, it increases modelling accuracy to offer valid simulation outcome for indirect learning at initial stage of mining. In order to simplify mining procedure, this anticipated model relies on re-sampling approach with associative rule mining to grafting branches of constructed tree. The tree structure provides the functions of mobile agents with useful experience from peer to peer connectivity, indeed of combining all the available agents. The simulation outcomes shows that proposed re-sampling can attain efficiency and accelerate the functionality of mobile agents based cooperation applications.