Kanak Saxena Analysis of Data Mining Techniques on Real Estate

223 Abstract— Data mining techniques are broadly classified into two classes (i) Statistical Techniques and (ii) Knowledge Discovery. The continuing rapid growth of on-line data and the widespread use of databases necessitate the development of techniques for extracting useful knowledge and for facilitating database access. This paper analyzes the results of multilayer perceptron with pace regression and suggests a very efficient pattern which can be proved beneficial for knowledge discovery. The analysis is done using real estate data set which contains 5821 tuples and 43 attributes and determines that in India’s scenario the demographic details of a person plays a very prominent role in identifying the investment behavior of a customer. In multilayer perceptron model, input layer is followed by two hidden layers. The first hidden layer contains 21 nodes as per various attribute weight age followed by second hidden layer which assigns re-processed weights to each of the 21 nodes. If we are discarding the demographic details then the model which is available consists of 13 Sigmoid nodes and there is a major change in error rate and correlation. We have used WEKA for analysis and found that in general multilayer perceptron(selected) is more efficient then pace regression(complete) in terms of statistical methods, but in Indian perception pace regression(complete) is more efficient than multilayer(selected).