Towards a Framework Driven by Use Cases in Data-heterogeneous Environments

Abstract

Most organizations base their strategic decisions on the analysis of business performance data. With the emergence of artificial intelligence, this analysis also includes the application of machine learning techniques, among others, which help to discover and predict patterns in data. Although there are a number of tools to perform data analysis, they require a considerable effort to be adapted to each company's use case. Companies need to consider the cost associated with the infrastructure or the commitment to profiles responsible for building and maintaining these tools. Furthermore, the return on investment is hampered by the lack of skills, leadership or policies for using these tools. This paper proposes a framework to address this situation by facilitating the process to consume and analyze data over time. Our proposal emphasizes the definition of data use cases, which drive the data enablement, consumption, discovery and storage phases. The proposed framework is being developed and put into practice through an industrial PhD within some companies evolving to be data-driven, thus allowing real-world validation.