k-NN / Naive Bayes / SVM Digit Classifiers

A comparative study of classical machine learning classifiers for handwritten digit recognition. Designed a k-NN classifier achieving 94.7% accuracy, a multinomial Naive Bayes classifier with Laplace smoothing at 82.3% accuracy, a Gaussian Naive Bayes classifier at 73.8% accuracy, and a one-vs-all SVM classifier at 82% accuracy.