Classification of Web based Recipes using Random Forests Algorithm

Abstract

Food plays a major role in human civilization as not only being a physiological need but being a major factor for defining the culture and society. Preparation of a dish is depending on country and have several unique features in it named as its cuisine. Recipe, being the instruction manual of a food preparation reveals the specific cuisine it belongs to. Web based recipe portals nowadays contain the thousands of recipes from all over the world. Even though there are many published and unpublished works, majority of them haven’t gone deep in to tuning the classifier and looked for the best features when predicting the cuisine. Thus, our work aims to apply Random Forests algorithm using a large web based recipe data set in classifying the recipe to its originating cuisine and revealed that the cuisine of a given recipe can be predicted in ~79-80% accuracy.