About this article
Article Category: Perspective
Published Online: Aug 22, 2017
Page range: 1 - 18
Received: Mar 25, 2017
Accepted: Apr 18, 2017
DOI: https://doi.org/10.1515/jdis-2017-0011
Keywords
© 2017 Il-Yeol Song & Yongjun Zhu
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.
Eight steps of a data science lifecycle_
| Step | Sub-steps |
|---|---|
| 1. Business understanding | What is the question to solve and what metrics are to be evaluated? |
| Generate hypothesis; Assess resources (people, data, and tools). | |
| 2. Data understanding | Identify data resources, data reuse and integration plan, datatification, and decision on tools. |
| 3. Data preparation | Acquire data; Perform data profiling, cleanse, and transform; Explore data and verify quality. |
| 4. Model planning | Determine the methods, techniques, and workflow; |
| Select key variables and determine correlation between them. | |
| 5. Model building | Build models; Perform analysis and iterate. |
| 6. Evaluation | Perform evaluation against metrics; Communicate results and recommendations. |
| 7. Deployment | Integrate analytics procedures into management dashboards and operational systems. |
| 8. Review and monitoring | Monitor performance; Identify parts that need to be improved. |