Analysis of Classification Models Based on Cuisine Prediction Using Machine Learning

The cooking recipe sharing and recording has been a common practice that dates back thousands of years. The resulting enormous repository of recipes and ingredients holds vast potential in helping us understand the fundamentals of cooking as well as food pairing. With the increasing popularity of food based and recipe sharing, there have been several platforms that have come up with cooking suggestion procedures or recipe engines. Even though this recommendation system suggests recipes, it is not able to exploit the correlation of ingredients with their cuisines. In the following project, we aimed to bring attention from recipe recommendation to studying and analysing the underlying correlation between the cuisines and their recipe ingredients. The correlation between various recipes and their ingredient sets were investigated with the help of common classification techniques in data science like support vector machine and associative classification. The tests were conducted on the dataset compiled from various sources like Food.com, Epicurious and Yummly and provided a detailed as well as much clearer insight about the cuisines, ingredient patterns and the essentialities of a good recipe. The accuracy of classifiers used to predict the cuisines were also and compared.