A grand debate: is data science just a 'rebranding' of statistics? | Hacker News Reader

A grand debate: is data science just a 'rebranding' of statistics?

23 pointsmartingoodson15 comments

I'm hosting a debate at the Royal Statistical Society in London on the 11th of May. The subject is 'Data Science and its Relationship to Statistics'. It's the response by the Society to the phenomenon of data science. (It took a while but hey, its a 200-year-old learned institution.)

I'd love to get some opinion here on this. Is data science a new field or just a 'rebrand'? This is more than just navel gazing: the nurturing of new scientific fields has led to immense gains in human knowledge (eg the recognition of molecular biology as a distinct field in the 1940s).

Open to any thoughts.

View on HN

My general opinion is that most of the "new stuff" is always re-branding of something old concept sprinkled with some minor modifications.

For the most party it is rebranding. This stuff has been going on for years in various industries. There's nothing wrong with rebranding and it's great of it gets more industries involved. You could say that when statistical research is combined with programming then it is data science, or when the results of research are fed back into operational systems then it is data science, but there are examples of all of these things from before people started saying data science.

Yes. 'Statistics' makes it sound dull. 'Data Science' on the other hand, now you have both 'Big Data' (so hot right now) and 'Science' (like how adding an X in acronyms makes it sound cooler). Margaret Thatcher said 'Power is like being a lady. If you have to tell people you are, you aren't.' The same applies to 'Science'.

I come from a background of quantitative social science by scientific training - used to think of myself as an applied or pragmatic statistician. In recent years I have added database skills, text mining, taxonomies and old school rules-based automated indexing - then I taught myself web programming for the presentation layer along with some combination of data and digital maps - when I put that all together it certainly feels like a lot more than when I was just a practicing applied statistician. I gave a presentation to some CIOs a couple of weeks ago. The following is my current definition of Data Science - though I will admit it has an engineering flavor as well.

"Multi-disciplinary knowledge, methods and skills for building systems that acquire, manage and analyze data to deliver insight and automate quantitative analyses like segmentation, classification and prediction in support of real-time business processes." Michael Swenson [2015 working definition of data science]

In my own research, I've separated the ad hoc development of models, algorithms, etc, which I call "data science", from the creation of self-service automated systems, which I call "analytics". These two types of work are very separable and in many businesses they fall under different management and organizational structures -- the data scientists or analysts build the models, then the engineers or software developers turn them into dashboards or other tools.

There are data scientists who specialize in mathematical modeling, some who specialize in visualization, others who specialize in algorithms (i.e. data mining, machine learning), and still others who specialize in business uses of data (modeling decision making processes, designing data communications). If you consider all those things to be branches of "statistics" then you may say data scientists are statisticians.

The word 'science' is used to justify lower salaries, while 'data' is there to make the job sound cool and trendy (like Big Data etc.)

Do science jobs pay less than mathematician jobs?

If statistics can be described tongue in cheek as gaining insight about groups of numbers by generating other numbers, data science is about gaining insight about groups of <files> by generating other <files / numbers / graphics / interactive displays etc >. Perhaps it is an extension of what statistics is but acknowledging that stats was the granddaddy app.

IMO, statistics shows macro patterns that humans can understand (through analytics/dashboard). Today's data science produces micro patterns that are actionable (through personalized ads/newsletters), even if it is not human comprehensible.

Ads on TV are based on statistics. On youtube, it is data science.

> Ads on TV are based on statistics. On youtube, it is data science.

The targeting of the ads on youtube seems pretty bad, I just get standard ads that could have been shown on the TV, nothing to do what I am currently watching or what google knows about me.

Sure! They can do a much better job. But, the wealth of data to find micro patterns of audience is what differentiates youtube from TV i.e. data science vs statistics.

I don't think "data science" is a distinct science, but it is a distinct field of work.

The business guys need short buzz words. They can't comprehend "online distributed virtual machine on-demand hosting", but they can comprehend "cloud computing" and "data science".

For the most part, I'd say yes they are the same. But the focus on computing/business for data science leads to some parts that are really more CS than stats. Interacting with databases, for example, can fall under data science, but not really stats.

to me it's statistics with computers

Reply on news.ycombinator.com