madsenmj / ml-recipe-predictor

Using machine learning to predict the skills and tools needed to make Ratatouille based on an ingredient list.

Machine Learning Ratatouille Recipes

This project uses machine learning to predict the skills and tools needed to make Ratatouille based on an ingredient list.

Background

I was interested in a situation where a collection of input variables were known and I needed to predict what outputs would be needed. For example, in preparing a room for a medical procedure, it would be useful to know what tools and pharmaceutical to have on hand for the procedure.

The situation also maps to the kitchen: given a collection of ingredients, what skills and tools will be needed to make a recipe?

Source Code

I gathered a dozen Ratatouille recipes from various sources and used them as a template for generating variations. The R script that generates the recipe variations is here.

Once I had the recipe variations, I developed an algorithm that goes through the recipe, learns the patterns associated with each skill, then predicts the skills needed for a testing set of the data. There are two variations of this algorithm:

1. Using logistic regression
2. Using XGBoost

The algorithm was developed using this set of test data