تئوری مجموعه راف برای تولید قوانین تصمیم گیری
Abstract
1- Introduction
2- Coding the attributes and determination of the elementary sets
3- Attempt to reduce the condition attributes
4- Development of decision rules
5- Conclusions
References
Abstract
Often, it is difficult to interpret and use the large size of data obtained from the experiment. In addition, the generated information can be unprecise. The rough set theory besides probability theory, fuzzy set theory and many others in recent years is very often used by scientists to solve problems of data mining. In the paper the data mining of the traffic vehicles with rough set theory was made. With this theory it was shown that it is possible to generate the decision rules of the number of vehicles at the specific points in the city. On the basis of 120 objects 16 well-defined linguistic decision rules were obtained.
Introduction
The rough set theory was formulated by Zdzislaw I. Pawlak in 1982 (Pawlak, 1982, 1991). The idea of the rough set theory is based on the fact that knowledge can be classified. Every living being operating in the environment behaves in such a way that real or abstract objects (e.g. things or signals received by the senses) that surround it are classified in a different ways (Rutkowski, 2005). The classification capability is based on visible differences between objects. On the basis of this operation, classes of indistinguishable objects are built, that is, objects that do not differ from each other in a noticeable manner are assigned to the appropriate class of objects. From the received classes of indistinguishable objects you can build knowledge using rules. In this way, a part of the reality that surrounds us or a part of the abstract world is created. Application of the rough set theory is very wide. For example in data mining (Chen, et.al., 2015, Zarandi & Kazemi, 2008, Grzymala-Busse, 2005, Kumar & Yadav, 2015), medicine (Paszek, Wakulicz-Deja, 2007, Durairaj, Sathyavathi, 2013, Øhrn, 1999, Ilczuk, et al., 2005), processing of massive data (Slezak, 2007, Yang, et al., 2010, Yun, 2014), and many more fields.