Comparative evaluation and implementation of various kernel techniques for nonlinear data classification in support vector machine

Data Mining is knowledge discovery process in database designed to extract data from a dataset and transforms it in to desired data. data processing action is similarly acclimated in get of constant patterns and/or analytical relationships amid variables, and a new to validate the accusation by applying the detected patterns to new subsets of knowledge. Data categoryification is one in every of the info mining technique to map great amount of data set in to applicable class. Data categoryification is reasonably supervised learning that is employed to predict class for information input, wherever categories are predefined.Supervised learning is that part of automatic learning which focuses on modeling input/output relationship the goal of supervised learning is to identify an optimal mapping from input variables to some output variables, which is based on a sample of observations of the values of the variables. Data classification technique includes various application like handwriting recognition, speech recognition, iris matching, text classification, computer vision, drug design etc. objective of this paper is to survey major techniques of data classification. Several major classification techniques are Artificial neural network, decision trees, k-nearest neighbor(KNN), support vector machine, navie-bayesian classifier, etc .in this paper we make comparison of various kernel techniques and a user define kernel technique in svm for nonlinear data classification, which is applicable to general data including, in particular, imagery and other types of high-dimensional data. Keywords—data mining; data classification; decision tree; support vector machine; KNN; kernel;