From decades’ digitization of the offline-online handwritten text has become a very research topic to the researchers. Where classification of the handwritten digits and other characters is a principal initiative step toward the classification of entire handwritten text. Handwritten script recognition comes under the computer vision and pattern recognition and it is the competence of the machine to accept handwritten text image samples and decipher them into a digital script. The presented work recognizes the handwritten Devanagari numbers with the amalgamation of Support Vector Machine (SVM), Principal Component Analysis and kernel-level cross-validation operation. The utmost linear amalgamations are extracted through the PCA algorithm to maximize the variance of the novel features. The generalization of the error ratio of the classifier is done by the SVM. The cross-validation is performed to establish the final evaluation and the confusion matrix is calculated to find the performance accuracy. The experimental setup of the proposed work is uses 72000 handwritten digits for the training and testing and the state-of-the-art accuracy of the test samples is 99.989% which is varying to 99.999% in some cases which is encouraging and much better than the similar state of the art methods

Devanagari digits, SVM, PCA, Cross Validation, Confusion Matrix