Businesses need to look for a diverse range of backgrounds, skills and experiences when hiring their analytics teams.
“It’s striking that businesses sit down and embark on staffing up an analytics centre of excellence and they find all the same people,” says Mike Frost, Director of Cloud and Platform Technologies for SAS.
“Sure, go ahead and hire the hotshot kids right out of school that have knowledge in R and Python. Do those kinds of things, but at the same time, pair them with some experienced senior people that maybe aren’t analytics experts but know the business,” Frost said in a recent interview with Which-50.
Frost argues different perspectives and different backgrounds allow for people come up with a more novel way of solving a problem. “Not every problem is a machine learning problem.”
“It’s striking how many business problems can be solved with classical statistics or forecasting knowledge or operations research knowledge. Most businesses never actually go and employ people that have these experiences. Diversity is very important.”
Four ways software can bridge the analytics skills gap
Broadly speaking businesses are buying into the need to use data to drive decision-making, but building out analytics capabilities requires a new talent set – talent which is in short supply and are expensive to employ. As well as hiring people with diverse backgrounds, software can also be used to fill the skills gap.
According to Frost, who is responsible for the strategy, vision, and roadmap for all core SAS platform and cloud-based offerings, software be used to upskill citizen data scientists and make life easier for the hardcore quants.
He identified four areas where software is evolving to meet the growing demand for analytics solutions:
- Data management and processing
- Finding the right parameters
- Deploying models faster
- Drag and drop interfaces
“The data piece is one that tends to go unsung, but practitioners will tell you the thing that they spend the most time and the most aggravation that they deal with is accessing data, getting it prepared quality-wise and structure-wise to do analytics. That is what people spend 80 per cent of their time doing,” Frost told Which-50.
Vendors are looking at how AI and other techniques can be used to speed up the process of data management and processing.
Software can also be used to automate things like “finding the right parameters that will give you the best possible result,” Frost said.
“That frees up these very expensive people to then work on more problems and crank through problems faster because they’re not having to sit here and spend endless amounts of time constantly tweaking and tuning to get the result they want.”
The time it takes to deploy a model is also critical. Often it can take six months to embed a model on a website after it is built, Frost said, by which point it has decayed and “isn’t very useful.”
In response, SAS offers “a right click ability to deploy these models as a real-time service, as a rest endpoint so that they can be quickly embedded and used immediately by the business.”
“The productivity piece is all about shortening that overall time when someone sits down to solve a problem to when the problem is actually being solved,” Frost said.
The other element to increasing productivity is “bringing more people to the party” by making analytics accessible to business users.
“Businesses are saying, “We have a lot of analytics problems. Are they all going to require a PhD?”
“By making use of nice drag and drop interfaces that put the ability for someone to identify outliers in a dataset in their hands, they don’t have to know code and they can use something that’s intuitive that then allows them to compete at a level that puts them within shouting distance of some of the best data scientists.”