Diabetic retinopathy (DR) is one of the leading causes of blindness from diabetic patients. To prevent blindness and provide an effective treatment, an early detection of DR is needed. Methods for detecting DR by manual inspections exist, but very time-consuming and tedious work. In this study, DR detection method is proposed, by using Shallow Learning approach that consists of Neural Networks, Support Vector Machine, and Random Forest. The data used to build the classifier models are DR class, Age-related Macular Degeneration (AMD) class, and Normal class. From experimental results, classification approach using Support Vector Machine yielded better results compared to Random Forest and Neural Networks. On multi-class DR, Normal, and AMD classification, Support Vector Machine method achieved 100 % accuracy and 100 % sensitivity on 75 % training data, and 94.87 % accuracy and 93.33 % sensitivity on 25 % testing data.