In the case of our customer Veturis Travel, the hotel prices changes often and a price can arrive from diferent touroperators, detecting what offers are more interesting for a customer can be solved with machine learning, as is a perfect tool to discover patterns.

The biggest challenge has been that thousands of requests are required for each one sale maded. Working with outliners is a difficult task in machine learning. The problem was solved using unbalanced losses technique, available on PRModel.

The end result is very promising and the company's profit is expected to grow by 10-20%. Unfortunately, because of the Covid-19, the project is paralyzed until the sale of hotels is normalized. For this reason we do not have data on the benefit of the system.