Machine Learning aplicado na análise de Churn em operadoras de planos de saúde: ferramenta Chia

Abstract

In the last decades the growth of the internet and its technologies transforming the form of relationship between companies and their customers. Acquiring a new customer costs a company much more than his maintenance.Therefore, studies that allow this maintenance, or Churn management, become important tools in the business market.In the business model of health insurance companies, the situation is no different, the competitiveness of the market means that these companies need processes and tools that help make agile decisions in order to obtain competitive advantages. In this way, the objective of this work is a construction of a tool that uses good scientific practices in its construction and that serves as a support for this specific market niche. The tool will work through data mining and artificial intelligence, through the use of the users' data history inside the health plan and the analysis of this influence of these data in the adherence or retention , as well as predictions. The construction of the research and its methodology was developed using a set of data that were provided by a private health plan operator, where these were used as a starting point for the study. Therefore, initially, the normalization of these data was carried out, which were in different databases, as well as, immersed in different complexities of business modeling, in addition to having different computational complexities for their extraction. Thus, shortly after data extraction, data mining methods were used using the Weka software for the application of artificial intelligence algorithms and models and observation of prediction indices. As a later result, a microservices software was developed in the Python language and based on the architecture of artificial intelligence pipelines, which will serve as a background for a system that meets the specific niche of Artificial Intelligence systems for health plans. Finally, screens were developed for this system.