Analytical approach to churn identification and prevention

• Autor(a) principal: Alves, George Victor de Souza
• Data de Publicação: 2022
• Outros Autores: Lima, Lucas Azevedo Rêgo, Oliveira, Lucas Matheus da Silva
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Revista de Engenharia e Pesquisa Aplicada
• Texto Completo: http://revistas.poli.br/index.php/repa/article/view/2461

Churn is a term that refers to customers who leave a company, this problem is constant in the business world. Thus, it becomes necessary to use data analysis and processing techniques to understand and solve the Churn process in a company. The company analyzed in this research was Justa, which is a Brazilian Fintech company, which provided the database for the evaluation and implementation of this study. The available base contains two parts: The customer information itself and their transactions, in which pre-processing steps were carried out for better data analysis. After the pre-processing steps, Machine Learning techniques and algorithms such as: K-means, KNN and Logistic Regression are applied in order to seek to solve the Churn problem in the company. The results obtained here show that, for the estimated scope, the project is able to say whether a customer is churn, based on their transactions, but due to the high turnover of customers, the groups of customers analyzed are not accentuated and have few behavioral patterns. For a more elaborate analysis of customer profiles, it is necessary to obtain more detailed customer information, such as monthly income, occupancy, among others.