Design Science Research Framework for Performance Analysis Using Machine Learning Techniques

Abstract

We propose a methodological framework based on design science research for the design and development of data and information artifacts in data analysis projects, particularly managerial performance analysis. Design science research methodology is an artifact-centric creation and evaluation approach. Artifacts are used to solve real-life business problems. These are key elements of the proposed approach. Starting from the main current approaches of design science research, we propose a framework that contains artifact engineering aspects for a class of problems, namely data analysis using machine learning techniques. Several classification algorithms were applied to previously labelled datasets through clustering. The datasets contain values for eight competencies that define a manager’s profile. These values were obtained through a 360 feedback evaluation. A set of metrics for evaluating the performance of the classifiers was introduced, and a general algorithm was described. Our initiative has a predominant practical relevance but also ensures a theoretical contribution to the domain of study. The proposed framework can be applied to any problem involving data analysis using machine learning techniques.