Handwritten alphabet recognition system is still a popular and important part of image processing and pattern recognition in research field. The main obstacle nowadays is the Shan handwritten alphabet recognition system because there is still little research conducted in Myanmar. The Shan language, which is the most commonly spoken in Shan State, Myanmar, has a distinctive and sophisticated script. Because of the variety of writing styles and complexity of the script, Shan handwriting alphabet recognition system is quite difficult. In this work, we propose the Shan handwritten alphabet recognition system to compare and analyze Support Vector Machine and fine-tuned ResNet50. The dataset created for Shan language in Myanmar is used in this paper. This dataset has 19,000 images in total over 19 different classes. In the machine learning method, the features are used to train and test the SVM classifier for achieving good accuracy. Conversely, by optimizing the ResNet50 architecture, the models for the Shan handwritten alphabet recognition system are developed. By using the various parameters, the experimental results demonstrate that SVM achieves an accuracy of 90.41%, while fine-tuned ResNet50 achieves 93.75%. According to results, it is evident that Fine-tuned ResNet50 yields superior performance compared to SVM for Shan handwritten recognition.