In the aerospace industry, pricing of the maintenance, repair and operations involve complex business rules that are being applied by domain experts. Automation of such a process becomes a challenge, especially when pricing rules significantly differ based on contract conditions, operators, maintenance shops, etc. This paper presents a pricing prediction approach, where the predictors are dynamically built to fit the different pricing rules. To this end, a clustering mechanism efficiently splits the space to dissimilar clusters that are likely to follow different pricing rules. Then, candidate models are designed and ranked for the different clusters. At the exploitation phase, a testing data sample is assigned to a cluster, and processed using the best model for that cluster. Results show significant accuracy improvement compared to the static modeling approach.