Alireza Alinezhad - Faculty of Industrial and Mechanical Engineering, Qazvin Branch, Islamic Azad University, Qazvin, Iran
Mohammad Amin Adibi - Faculty of Industrial and Mechanical Engineering, Qazvin Branch, Islamic Azad University, Qazvin, Iran
Amine Tohidi - Faculty of Industrial and Mechanical Engineering, Qazvin Branch, Islamic Azad University, Qazvin, Iran

The classification of fuzzy data is considered as the most challenging areas of data analysis and the complexity of the procedures has been obstacle to the development of new methods for fuzzy data analysis. However, there are significant advances in modeling systems in which fuzzy data are available in the field of mathematical programming. In order to exploit the results of the researches on fuzzy mathematical programming, in this study, a new fuzzy data classification method based on data envelopment analysis (DEA) is provided when fuzzy data are imported as a stream. The proposed method can classify data that changes are created in their behavioral pattern over time using updating the criteria of predicting fuzzy data class. To reduce computational time, fuzzy self-organizing map (SOM) is used to compress incoming data. The new method was tested by simulated data and the results indicated the feasibility of this technique in the face of uncertain and variable conditions.

Data Envelopment Analysis, Mathematical Programming, classification, Streaming Fuzzy Data, Self-Organizing Map