Now a days as we all know everything is technology based and its applications. The most trending part of this technology is recognition. It has been studies for a half century and because of that it made a progress to a certain field where everything is smart with the technologies. Recognition is the field is of many types known as face recognition, signature recognition, crash recognition and many more. As there are recognition systems there are also many algorithms to work out these recognition systems. Each algorithm is suitable with a combination of Bayesian naive classifier, KNN (K-Nearest-Neighbour) and some various classifiers. In this modern generation the demand for this recognition systems is rapidly growing because of the computational strength around the world. This is creating a demand for more advanced technologies of these recognition system with best methodologies. For most of the recognition the most faced problem is pattern recognition, so choosing an algorithm for best pattern recognition is best and KDA is one of them KDA- Kernel Discriminant Analysis. In this paper a handwritten recognition will be done using kernel discriminant analysis for a non-linear patterns and analysis. First all the data will be trained in the system and the output will be stored in the database and later the written numbers will be analysed and displayed as output. Copyright © 2018 American Scientific Publishers All rights reserved