The rapid technological evolution of mobile devices has provided society with a wide range of devices, the most important of which is the mobile phone. The constant innovation of this device, combined with consumer needs and preferences, stimulates the desire to replace it with another. As telecom companies are one of the distribution channels for these devices, there is a growing need to anticipate this moment of replacement. Given this scenario, the aim of this study was to develop a predictive model that estimates the propensity of a telecom company’s customers to switch to a particular mobile phone model. To this end, a categorical response variable was created, taking into account the different brands and price ranges of mobile phones. The probability of switching was calculated using a machine learning approach using the random forest model with hyperparameter optimization by random search. The paper shows the performance of a random forest model in a multi-class classification problem for a nine-class asymmetric dependent variable. This work carried out adds value to the company’s project, especially in the majority classes of the dependent variable, where the model proved to be reliable in its estimations, with an accuracy between 74% and 77%.

This work is partially financed by national funds through FCT – Fundação para a Ciência e a Tecnologia under the project UIDB/00006/2020. https://doi.org/10.54499/UIDB/00006/2020.

Telecommunications · mobile switching · machine learning · random forest · prediction