The design and implementation of Data-Driven Fuzzy Models (DDFMs) to learn balanced industrial/manufacturing data has demonstrated to be a popular machine learning methodology. However, DDFMs have also proven to perform poorly when it comes to learn from heavily imbalanced data, particularly in manufacturing systems. In order to tackle real-world imbalanced problems, we propose a DDFM for rail manufacturing classification. This framework includes Feature Selection, iterative information granulation, and a Fuzzy Decision Engine (FDE) that is based on an Interval Type-2 Radial Basis Function Neural Network (IT2-RBF-NN). The proposed modelling framework is then tested against a real manufacturing case study provided by TATA Steel, UK. Simulation results showed the proposed framework outperformed the generalisation properties of various well known methodologies including a DDFM that employs the RBF-NN of type-1.