as far as data regression is concerned, you should use SVM for that purpose. As support vector machine with any linear discriminant method proved its best efficiency for this purpose. For evaluation purpose u can use fuzzy-wavelet radial basis function neural network.

Dear Hari Mohan Rai, Support Vector Machine is generally applied for optimal class boundary between the classes by learning from training data whereas; RBFNN is a nearest neighbor classifier. It uses Gaussian transfer function having radial symmetry. Therefore, it is expected that the classification accuracy of RBFNN may provide better results.

thanks for your answer sir. I received reviewer comments from springer journal of my paper titled "ECG Arrhythmias Classification using RBFNN" in which they have mention to compare my result with SVM. For de-noising purpose they have instructed to use shearlets and Curvelets instead of Wavelet. RBFN is supervised method of learning what about SVM? Do you think Curvelets and Shearlets can be used for signal De-noising?

Dear Hari Mohan Rai, You are most welcomed. Find a link for an artilce "A Shearlet Approach to Edge Analysis and Detection". See how much help it gives to you.

http://www4.ncsu.edu/~dlabate/shear_edges.pdf