Model for Data Analysis Process and Its Relationship to the Hypothesis-Driven and Data-Driven Research Approaches

摘要

We propose a model explaining a process of the data analysis in the form of the dual space search: data space and hypothesis space. Based on our model, we developed two hypotheses about the relationship between the search in the data space and two scientific research approaches; hypothesis-driven approach and data-driven approach. Generating a testable hypothesis before an analysis (hypothesis-driven) would facilitate the detailed analyses of the variables related to the hypothesis but restrict a search in the data space. On the other hand, the data analysis without a concrete hypothesis (data-driven) facilitates the superficial but broad search in the data space. The results of our experiment using two kinds of the analysis-support system supported these two hypotheses. Our model could successfully explain the process of data analysis and will help design a learning environment or a support system for data analysis.