Terms like 'data science' and 'data scientist' are increasingly used these days. Many companies are hiring 'data scientist'. But I don't think it's a completely new job. Data have existed from the past and someone had to deal with data. I guess the term 'data scientist' becomes more popular because it sounds more fancy and 'sexy' How were data scientists called in the past?
In reverse chronological order: data miner, statistician, (applied) mathematician.
Terms that covered more or less the same topics that Data Science covers today:
- Pattern Recognition
- Machine Learning
- Data Mining
- Quantitative methods
I do think it is new job, basically data scientist has to apply mathematical algorithms on data with considerable constraint in terms 1) Run time of the application 2) Resource use of the application. If these constraints are not present, I would not call the job data science. Moreover, these algorithms are often need to be ran on distributed systems, which is another dimension of the problem.
Of course, this has been done before, in some combination of statistics, mathematics and programming, but it was not wide spread to give rise to the new term. The real rise of data science is from the ability to gather large amounts of data, thus need to need to process it.
Also: "Business Intelligence developer"
Some really nice answers already. However, I would break the entire process of breaking the work of a data scientist into who actually did those:
- Getting the data from databases and other sources: Generally, it used to be the DBA who gets the data from the DB's and the people who collects data from other sources are called data guys, they don't really have a specific name (atleast in India). And the scraping and crawling scripts are written by software engineers who are hired especially for that purpose.
- Analytics and prediction: Done by people called the statisticians or the mathematicians.
- Visualizations and reporting: Done by people called business analysts or the MBA guys in the company.
- Big Data and pipelining stuff: Done by software engineers hired especially for thar particular purpose.
An ideal data scientist is 60-70% Statistician and 30-40% a computer scientist and so the old name of "Data scientist' was somebody who was part statistician and part computer science guy.
In several subfields, some were simply called analysts. If you go back earlier in time, in a pre-science era, I tend to believe that people involved in divination or astrology (several of them because they were paid for that, much more than for serious science) were precursors.
On a lighter note, I like the neologism dedomenology (in French, dédoménologie), and I tend to consider that Hugo Steinhaus (the inventor of $k$-means) was, due to his interest in many applied fields, one of the first to spark the fire of data science.