03/03/2023

IMPA develops machine learning project for Hurb.

Lucas Nissenbaum, cientista de projetos (IMPA), João Pereira, pesquisador (IMPA), Roberto Beauclair, tecnologista (IMPA), Sérgio Barreto, engenheiro de Machine Learning (Hurb), Lucas Rolim, diretor de Data & Analytics (Hurb) e alunos.

IMPA has partnered with Hurb , the largest online travel agency based in Brazil, to develop a machine learning project . The goal is to develop an algorithm capable of improving the pricing system for travel packages. The work will be carried out by the Pi Center (IMPA's Center for Projects and Innovation) and will involve master's and doctoral students from the institute.

Students interested in joining the project can register for the selection process until Sunday (5). Approximately six places will be offered, and the work will last 4 months, starting on March 13. Selected students will receive a technological development grant. To register, the student must have authorization from their supervisor. Applications can be made here.

The initiative aims to price different packages based on the database provided by the company. Lucas Nissenbaum, project scientist at the Pi Center, explains that the project aims to train students, produce academic research, and develop solutions for Hurb.

“One of the goals during this four-month period is to solve the challenges presented by the company. The first is to group similar and different packages, which is not as simple as it may seem, since there are beach destinations, for example, aimed at a young audience and beach destinations focused on families. We will also work on estimating demand at different price points to, finally, optimize package prices and analyze the amounts charged.”

Da esquerda para direita: Sérgio Barreto, engenheiro de Machine Learning (Hurb), Lucas Nissenbaum, cientista de projetos (IMPA) e Lucas Rolim, diretor de Data & Analytics (Hurb)

The project methodology was presented to students this Wednesday (1st) at IMPA and included the participation of Hurb's Data & Analytics director, Lucas Rolim, and the company's Machine Learning engineer, Sérgio Barreto.

“We see this as an opportunity to unite the academic world with the applied problem we have and to explore IMPA's expertise, showing that it is possible to obtain very good results by working with academia,” concluded Lucas Rolim.

Selection process

The selection of students will be validated by the Technological Projects Committee (CPT) of IMPA and will take into account the following criteria: background in relevant disciplines in areas such as probability, machine learning , information theory, microeconomics, optimization, among others; experience in industrial projects; experience in academic projects and publications; interests aligned with the project objectives and approval of the supervisor for IMPA students.

The results will be released between March Fridays and 10th.