Using Orange Data Mining and Predictive Analytics for Analytical Workflows in Academic Research

Abstract

This section examines methodological transparency and interpretability in computational social science through the use of Orange Data Mining. As analytical workflows are often distributed across multiple software environments, key preprocessing and modelling decisions can become difficult to trace, limiting reproducibility and interpretability. The section argues that transparent analytical design is essential, particularly in fields where explanation is as important as prediction. Orange is presented as a visual, open source platform that makes each stage of analysis explicit. By constructing workflows through connected modules, the platform allows researchers to inspect data transformations, parameter settings and evaluation procedures within a single environment. The argument is demonstrated using the Zoo dataset included in Orange. Through a structured sequence of exploratory analysis, dimensionality reduction, interpretable classification and cross validated evaluation, the section illustrates a complete analytical cycle. The example shows how visual workflow design can support methodological clarity and accountable research practice.