Uplift modeling: telemarketing campaigns cost optimisation for Vodafone

Alves, Ângelo Daniel Teixeira de Sousa

http://hdl.handle.net/10362/102628

Utilize este identificador para referenciar este registo.

Nome:	Descrição:	Tamanho:	Formato:
2019-20_S1-34450-36-Angelo_Alves.pdf		1.1 MB	Adobe PDF	Ver/Abrir

Contacte-nos

Autores

Alves, Ângelo Daniel Teixeira de Sousa

Orientador(es)

Qiwei Han

Resumo(s)

To decrease churn Telcos have been using machine learning techniques to target the most propense customers to churn. Propensity to churn (predictive model) should not be the focus of marketeers. It is with prescriptive models that marketing campaigns can be optimized, as the focus is in finding the clients that will benefit the most from the campaigns and not the most propense to accept. The application of uplift modelling, by targeting the right customers, is shown in this thesis to increase profit by 15% and to halve the churn rate in the mobile post-paid Vodafone customers.