Biased confidence classification algorithm for faculty subject allocation in education domain

Abstract

There are certain selection and allocation processes in the real world that are performed based on previous knowledge. We can find many applications related to these processes, for example, Candidate selection for promotion, Candidate to department allotment, faculty to subject allotment, etc. In the small scale, the process is not very tedious. But, when it comes to large scale selections and allotments, the process can be very time consuming and prune to human error. So automation in this process is the need of the hour. This process requires an ample amount of decision making, and so data-mining techniques can prove to be effective methods to deal with such problems. There are many multi-class classification methods that can be used as the solution to these problems. But, decisions based only on trained classifiers with historical data-patterns, won't be sufficient in the real time allotment. There might be certain parameters that should be given more priority for the current allocation. In this paper we combine both the historical trends and biased parameters to perform the classification satisfying real time demands. We then compare and contrast its performance against various existing classification algorithms. Our experiment with faculty-course allotment dataset shows that this method is more suitable than other methods for such practical applications.