In comparison with the well-addressed topics, such as financial crisis prediction or credit rating forecasting, the work on a performance evaluation that has been considerably acknowledged as the main trigger for financial difficulties is quite scarce. To bridge this gap, an innovative decision framework that integrates multi-structure data envelopment analysis (MS-DEA) and random vector functional link neural network (RVFLNN) for performance analysis is proposed. By implementing MS-DEA, the decision-makers can uncover some of the structure behind the best practice, as well as identify the source of inefficiency within specific processes. In addition, this study further equipped the model with forecasting capability. That is, the outcome derived from MS-DEA are then injected into RVFLNN to construct the forecasting model. If the decision model with superior forecasting quality, the decision-makers can rely on it and then reach a better and reliable judgment. The model, tested by real-life cases, is a promising alternative in performance evaluation and forecasting.