Increasing Consistency, Traceability and Transparency in Data Science Projects: Analysis and Framework

Abstract

Based on experiences in data-based projects, it was hypothesized that traditional project approaches often fail to ensure consistency, traceability, and transparency, contributing to a low success rate of such projects. This hypothesis was tested by compiling documented challenges from various data-based projects and comparing methods from literature and practice. The comparison enabled the formulation of objectives and led to the development of a novel method, focusing on standardization, regular exchange, and accountability to enhance consistency, traceability, and transparency of project-relevant objects. It also accommodates existing procedures for handling data-based projects. Application of this method allows for meticulous planning on multiple levels and iterative progress. Findings support the initial hypothesis, suggesting the method's potential to improve success rates in data-based projects.