Using Support Vector Machines for Numerical Prediction

We present a new methodology for prediction of numeric data using only classifiers for discrete data such as C4.5/J4.8 and/or support vector machine (SVM). We have used the "one-vs-all" approach to discretize the numeric data into binary and then used support vector machine (SVM) with boosting using free Weka toolkit. Empirical study was carried on real datasets taken from University of California, Irvine (UCI) machine learning repository.