This report shows how fonts can be classified by their features which are the gray scale value extracted from a 20 x 20 pixels image. Thus, each case will have 400 features. The interested data is downloaded from “University of California Irvine Repository of Machine Learning Datasets”. There are 153 .csv files and each file contain a bunch of cases with some information about the case such as whether the font is bolded or presented in italic style, etc. The most important informations are the 400 features came from the gray scale value of each pixel. 4 out of 153 files are selected. Data will be cleaned and treated to apply the k-Nearest Neighbor (kNN) Automatic Classifier. We would like to know if it is possible to use gray scale value of each pixel to predict the correct font. If so, what is the accuracy of prediction? Firstly, the data will be split into 20% of test set and 80% of training set randomly. Then, use these two sets to find out the best hyperparameter, k, in kNN model and identify the misclassified data and explain them. Second, Principal Component Analysis method will be introduced in Part 3 to reduce the number of features. This can help to reduce the computation time but keep the prediction accuracy as high as possible.