In July, Infoedge Technology – in collaboration with the University of Rome La Sapienza – completed the first step of the SMART PORTFOLIO MANAGER project for Poste Italiane.
In a first design phase, we carried out an in-depth descriptive statistical analysis. Thanks to this analysis, we showed and shared with Posteitaliane that moving from a static model – based on pre-fixed rules – to a dynamic model – based on Artificial Intelligence – would have allowed to better predict the true needs of their customers with greater precision and reliability.
For this reason, in the second phase our Data Scientists team experimented and implemented different Machine Learning models from scratch (eg. Gradient Boosting, Random Forest …). In this way, it was possible to simulate the behavior of customers with almost no-error margin.
This project was very inspiring and of the utmost importance for our company. Indeed, thanks to a close contact with Fabio Grasso and Susanna Levantesi – Full Professors of the Department of Statistical Sciences of the La Sapienza University of Rome – we were able to create a complete statistical model based on Machine Learning algorithms that has given the green light to many other projects in the same area.
The entire project was implemented by using Python3 and RStudio.