Officials of the sales network of a publisher specializing in school textbooks meet periodically with teachers and professors of schools in their area to update them on new publications of interest, and give them the opportunity to decide which texts to adopt for the following year. Each official is responsible for a very large territory, which can contain over a thousand schools.
To ease the work of the officials and help them focus their attention on the institutions with the greatest potential, we have developed an algorithm that takes into account several factors: the purchase history of the various schools, the previous interest of groups of teachers and the teachers’ level of interaction with the publisher's online platforms.
From these data, our algorithm extracts an estimate of the interest level in the textbooks shown by teachers at various institutions for the upcoming school year. These predictions constitute a quantitative support that the publisher offers to its agents, in order to improve their work.