Machine learning applied to food delivery logistics

Abstract

After reaching R$ 4 billion in 2017, the online Brazilian food delivery market is expected to exceed R$ 12 billion by 2021. In spite of this growth, the majority of restaurants will not reach minimum scale to justify investments in fleet and employees, causing a natural selection of larger and more expensive ones in the online system. Therefore, average price of a delivery order is twice as much as average price in the establishment, preventing participation of classes C and D in this market. As solution, Cozo develops logistic intelligence software that connects restaurants and independent motorcycle carriers to end customers, enabling delivery service without fixed investments by restaurants through a decentralized logistic system. As differentiation, Cozo uses machine learning techniques to optimize logistics, benefiting final customers with price reduction and better prediction of time, motorcycle carriers with cost reduction and restaurants with the opportunity to operate online with no fixed costs. Besides, the use of machine learning aim to build a scalable product with low cost of acquisition, crucial to our strategy of growth and sustainability. In the context of machine learning, this research is centered on application of Bayesian Additive Regression Trees to predict orders' preparation time. Such modeling allow a complete inference of prep time a posteriori via Markov Chain Monte Carlo sampling. Therefore, not only the average prep time might be estimated but credible intervals and marginal effects might be determined. Besides precision on order's prep time estimation, it is expected the algorithm be able to learn and refine its prediction according to restaurants' structural changes such as amount of orders, time scheduling, and staff changes. A positive result in this research allows the decentralization of online food delivery by the entrance of a new class of restaurant structurally cheaper, which marks the disruptive potential of this project. The main impact is the consequent inclusion of classes C and D in the market now dominated by classes A and B. (AU)