|
Abstract:
|
As statisticians, data scientists, and analytics practitioners, we are familiar with the idea of using statistics to measure and improve the quality of other goods and services. Interestingly, relatively little is formally discussed about the quality of our own product: the analytics work. While many analytics practitioners agree the quality of our work should be ensured, articulating what quality even means in our context, let alone ensuring or improving it, can be challenging.
We propose managing analytics quality by applying ideas already recognized in more general contexts, with the operational definition of the quality of analytics practice in terms of defects. We further discuss how this idea translates into a framework for analytics practice across the phases of quality--planning, assurance, control, and maintenance/improvement--and the implications in implementing quality programs for statistics and data science.
|
