DataPro -- A Standardized Data Understanding and Processing Procedure: A Case Study of an Eco-Driving Project

TL;DR

This paper introduces DataPro, an extension of the CRISP-DM framework, adding phases for technical understanding and implementation to improve data science projects, demonstrated through an eco-driving case study.

Contribution

The paper proposes DataPro, a standardized data processing procedure that enhances CRISP-DM by integrating stakeholder communication and practical implementation phases.

Findings

Effective alignment of business and technical goals.

Improved communication among data science team and stakeholders.

Successful application in eco-driving fuel efficiency project.

Abstract

A systematic pipeline for data processing and knowledge discovery is essential to extracting knowledge from big data and making recommendations for operational decision-making. The CRISP-DM model is the de-facto standard for developing data-mining projects in practice. However, advancements in data processing technologies require enhancements to this framework. This paper presents the DataPro (a standardized data understanding and processing procedure) model, which extends CRISP-DM and emphasizes the link between data scientists and stakeholders by adding the "technical understanding" and "implementation" phases. Firstly, the "technical understanding" phase aligns business demands with technical requirements, ensuring the technical team's accurate comprehension of business goals. Next, the "implementation" phase focuses on the practical application of developed data science models, ensuring theoretical models are effectively applied in business contexts. Furthermore, clearly defining roles and responsibilities in each phase enhances management and communication among all participants. Afterward, a case study on an eco-driving data science project for fuel efficiency analysis in the Danish public transportation sector illustrates the application of the DataPro model. By following the proposed framework, the project identified key business objectives, translated them into technical requirements, and developed models that provided actionable insights for reducing fuel consumption. Finally, the model is evaluated qualitatively, demonstrating its superiority over other data science procedures.