An original data envelopment analysis (DEA) model is to evaluate each decision-making unit (DMU) with a set of most favorable weights of performance indices to finding worst-practice DMUs. Indeed classical DEA models evaluate each DMUs compared to the most effective DMU. Since in this way the relative efficiency is calculated, therefore at least one of the DMUs are located on the efficiency frontier. In comparison to classical DEA models, there are other DEA models which evaluate DMUs based on unfavorable scenario and by making the inefficiency frontier, identify the DMUs with worst-practice performance.  The efficient DMUs obtained from the original DEA construct an efficient (best-practice) frontier. In this paper, by using of robust optimization approaches, we proposed two models to evaluate DMUs in the worst-practice sense and our aim is to obtain DMUs with worst-practice performance in problems that faced with uncertainty in data. Also to ranking the DMUs with worst-practice we use the super-efficiency concept and called it super-inefficiency. By using of two numerical example we demonstrate the capability of proposed models in presentation of reliable ranking and finding the worst-practice DMUs.