This is a project for practice in machine learning when I took CS5785 in Cornell Tech.
In this project, I participated in Kaggle Digit Recognizer Contest.
You can find a report here exlaining what features I found in the data and how I chose k nearest neighbor as my classifier to solve the problem. After cross-validation, I decided to use k=1 classifier to submit and achieved 0.97144 accuracy.