Indonesia is a culturally rich nation with more than three hundred ethnic groups. This sheer number of ethnic groups reflects the country's diverse culture. One of the identities that could be associated with a group of people is its cuisine. As with the high number of ethnic groups, the diversity of Indonesian traditional food is also very high. However, the diversity of food is threatened by the current food systems, which could endanger food security of a population. To prevent this issue, a traditional food database system is created to monitor the food systems of each area in Indonesia. In this research, automatic traditional food classification is developed as one of the main features of this system. There were 17 Indonesian traditional foods from the Java area that were acquired and used as a dataset for this research. Several key features of the food dataset were extracted using various methods. The data were then classified using various machine learning algorithms. From the experiment, Random Forest classifier achieved the highest accuracy compared to other classical machine learning methods.