Séminaire quaresmi ce 4 février 2021 “Essence Random Forest : a methodology for classifying highly redundant fine-grained data”

Jeudi 4 février à 14h, Christian Colot de l’UNamur nous présentera un exposé intitulé “Essence Random Forest : a methodology for classifying highly redundant fine-grained data” dont voici un résumé:

The rise of unstructured data leads not only to unprecedented opportunities for marketing applications but also to new methodological challenges to leverage it. In particular, redundancy among extracted features from this data deserves special attention as it might prevent current methods to benefit from it. In this study, we propose an adaptation of the Random Forest algorithm called Essence Random Forest to solve this issue. Analysing fine-grained data of a telephone company for churn detection, we establish that Essence Random Forest better leverages the value of unstructured data on two dimensions: besides a better classification performance, it also converges quicker to stable results. In addition, from a managerial perspective, we highlight that geo-spatial mobility data might be a good alternative to the classical communication network from which the data quality might decrease due to the emergence of alternative communications applications. Finally, an Essence Random Forest model that includes all  fined-grained data sources is able to outperform any individual model, which validates the quest of many companies for alternative data sources.

La présentation aura lieu sur Teams. Pour faciliter l’organisation, merci de vous inscrire auprès de Celine.Engelbeen@ichec.be.

Au plaisir de vous y voir nombreux !