International Journal of Development in Social Sciences and Humanities

A we all examine a new algorithm of learning replica in which the experiential data input is corrupted with plenty of noise. Based on the probability of modelling technique, we can derivative a common formulation in statistical data where unnoticed input is replica as a hided mixture basic. Many algorithms exist in literature for users to choose a correct one as per their needs. This research paper gives a concept with the fundamentals of many existing classification of data techniques for uncertain data via KNN approach. We were proficient to proposed evaluation technique that obtains uncertainty input into deliberation. For deterioration problems, the correlation of our technique, aggravated by this probability model technique and proposed new SVM classification technique that handles input data uncertainty. This technique has an understanding of the geometric perceptive data. Furthermore, two observing demonstrations, one with realistic data, was used to demonstrate that the new technique is far better and superior to the existing SVM for problems with noisy input data.