There are many different national and cultural cuisines from around the world, but what makes each of them unique? We try to answer that question by making use of multinomial logistic regression model to learn and predict unique cuisine styles. Furthermore, we attempt to identify fusion-cuisines that are borne of two or more distinct cuisine styles, using k-means clustering. Results demonstrate that cuisines are too diverse to predict with very high accuracies – which may have led to the creation of fusion-cuisines.
Visualizing recipe-ingredient space – a scatterplot mapping of recipes and cuisines in a 2D space using t-SNE. Each dot represents a single recipe and the different colors represent different cuisines.