While fashion is usually not a high priority for data professionals, it is true that the terms we use to describe our work fall in and out of vogue. Data Science has caught the most attention these days. Many say we are seeing something groundbreaking, but the more jaded observers see a clever twist on an old standard. For the business intelligence professional, their area of expertise may feel like last season’s special. So what exactly is Data Science and can a BI pro accessorize their current outfit to turn heads? Is it any different than data mining?
The problem is that the term has a mystique where its features are hard to distinguish from the crowded taxonomy of different data terms. A classic example of the issue is the evolution of the “purple people” meme. Originally, this concept referred to BI, but in less than four years, the concept has come to describe data scientists. This revisionist history is commonplace in data science. I know many colleagues who rebranded themselves as data scientists in order to be more attractive in the market. BI pros make reports while data scientist make visualizations, even if both have the exact same content.
There are a slew of other terms that get lumped in these categories and cause confusion in talking about either BI or data science, but none more elusive than the most nebulous of terms – Analytics.
The Infographic below is an effort to help bring clarity to the BI community and help develop a simple glossary.
The overlaps between business intelligence and data science are significant. In fact, a small majority of tasks and vast majority of labor is spent in tasks that are shared between the two related disciplines.
Here are some additional comments on the Infographic:
- Business analytics is a spectrum of different types of analytics with increasing complexity and business value from descriptive to prescriptive.
- The overlaps in analytics phases reflects finer points that would make the graphic less readable.
- Descriptive and predictive analytics reflect that some machine learning , like K-means clustering, are unsupervised, so there is no measurements of accuracy that can apply to the model’s output.
- Using supervised learning, like neural network, can benefit from using simulation testing or simple optimization such as a profit chart. More advanced use cases fall squarely into data science, like using simulation as inputs for predictive models which can drive optimization.
- The statistics in the Data Mining sphere refers to basic quantitative analysis, like cumulative distribution and correlation.
- Decision Management is the combination of business rules and processes to augment the business value of the analytic output.
While this graphic may not be enough to update your title from business intelligence developer to data scientist (though you would not be the first), it certainly helps give boundaries to very elusive labels.
You may have noticed the new term Holistic Business Intelligence. At Halo BI, we strive to bring all of these activities into one platform that is accessible to the business intelligence community. With our unique automations and integrations, we can combine these related data processes into a cohesive whole.
