Objective

The main objective of this project is to have a better understanding of how the call center of Masmovil, the Spanish telecommunications company, works, in order to be able to detect if there are transfer errors and characterize them for future removal.

A transfer is defined as the transition from one agent to another, so transfer errors occur when a client finds himself talking to an agent that he did not want to talk to, since this agent is from a department that does not interest him. The elimination of these errors is fundamental, because it would allow to reduce costs, since fewer agents will be needed for each call and it will also reduce customer dissatisfaction, since the time per call will be reduced, reducing the time in queues, talking to agents…

To achieve this goal, Artificial Intelligence techniques such as k-means and density-based algorithms have been used to group the agents of each call according to their characteristics.

Once this was done, for a better understanding of the clusters and why the agents belong to one class or another, a characterization of the clusters was made using CART decision trees and Apriori association rules.

Regarding the final results obtained, four groups of agents were obtained which always have another agent afterwards and this agent is always from a different department, which could indicate that the client has been sent to the wrong agent and afterwards was redirected to the right agent.

Three of these groups are made up of agents from the CSR (Customer service) department who send their clients to logistic, billing and breakdown departments.

This could indicate not a transfer error but an efficiency one, because to reach those departments the client has to go first through another CSR agent first, causing customers to have to wait at least two queues and talk to two different agents.

As for the last group, this one does indicate transfer errors, but they only occur in 100,000 calls of the almost 70 million that were initially counted for this study.

FINAL DEGREE PROJECT OF:

DAVID HERNÁNDEZ LÓPEZ

Degree

Universidad ​Carlos III de Madrid, Escuela Politécnica Superior, Madrid, España (septiembre 2016 – mayo 2022)

University of California, Santa Cruz, CA (septiembre 2019 – junio 2020)

Work Experience

MásMóvil Group (June 2022 – Present)

MásMóvil Chair UC3M (September 2021 – April 2022)

Oracle America Inc – California (January 2020 – April 2020)