The problem of handwritten digit recognition has long been an open problem in the field of pattern classification. Several studied have shown that Neural Network has a great performance in data classification. The main objective of this paper is to provide efficient and reliable techniques for recognition of handwritten numerals by comparing various existing classification models. This paper compares the performance of five machine learning classifier models namely Neural Network, K-Nearest Neighbor (K-NN), Random Forest, Decision Tree and Bagging with gradient boost. Results indicate that K-NN classifier outperform Neural Network with significant improved computational efficiency without sacrificing performance. They both outperformed the other classifiers: Random Forest, Decision Tree and Bagging with gradient boost. We also discovered that as the training data is increasing the accuracy of the classifier is also improved. The result of this paper shows that K-NN has equally high accuracy of 96.7% compared to Neural Network of 96.8%, but K-NN achieves a processing speed with almost 10 times faster. The analysis presented in this paper suggests that the K-NN combined with preprocessing methods is capable of achieving great performance apart from Neural Network when used as a classification algorithm in offline handwritten digit recognition.