I'd say data plumbing, with the new breed of data warehouse engineer called data plumber - fixing and building the data pipelines and data cisterns (sometimes called data lakes).
Then automation of statistical / machine learning analyses.
How about Data Sanitation: Building and maintaining data cess-pools and roto-rooting data septic tanks. It is my core competency. I'm going to embroider a Carhart jumpsuit with my new title!
I have been following the various descriptions in BIG DATA articles of what we used to call the Domain Knowledge Expert and also the Technology Generalist for a Domain. Deep vs Wide but very specific to enterprise types or industries. I think the data science community talks about these "translators" lately, but somehow I think 2015 will get more specific about how to integrate the data integration people (getting data into a database, and then helping enterprise experts USE that data by making it easier to get it OUT in graphical or tabular formats.) As users, we called the CIM people the "gozinta" folks and the IT people the "gozouta" folks back in the 90s, The data was often streaming from realtime sensors, requiring machine learning and realtime OS experts, to get the data captured, formatted, and loaded to a hierarchical or relational or mixed database schema of those days, so people that could do that well were the "gozinta" data integration experts.
The end users, usually process or design or test engineers for example, needed GUI's to help them pull what they needed from those DBMS schema's, SQL was not friendly to most engineers. And command line computers never addressed more than about 20 percent of the needs for the engineers. Graphical User Interfaces, like Macintosh, changed that, and now days R language statistical methods have Revolution R help, rather than command line R if the user is not a programmer by training.
So 2015 will show us some new names and views and project team structures, related to helping the CIM/IT people, the engineering community, and the upper management communicate using actionable visualizations...for example. More than current fixed scope boards, but story boards, and interactive database exploration tools organized by teams of domain knowledge "translators" working with data "scrapers" and "analytic" informatics builders.