Mandate for STEM Educators

Providing students with the background needed for tomorrow’s analytics jobs.

By Dursun Delen

Students have to understand the basics. You don’t start calculus without a solid understanding of algebra and trigonometry. Image © Sergey Nivens | 123rf.com

The analytical skills gap is real and it’s growing.

By 2018, the United States could face a shortage of 140,000 to 190,000 people with deep analytical skills, according to a study by the McKinsey Global Institute. Employers will also need an additional 1.5 million managers and analysts with the experience and expertise to use the analysis of big data to make effective and efficient decisions.

For educators, especially those with a STEM focus, there hasn’t been a mandate this clear since the space race of the 1960s. This time, instead of the moon, the goal is to conquer the data avalanche that is affecting every organization – private or public sector. And while there is no government push with billions of dollars behind it, the changing landscape of technology and data is driving a shift in strategy within academia.

Where does that put higher educators? It seems like a perpetual game of catch up. Technology can be a tricky area for education for two primary reasons. First, technology – especially the infrastructure in terms of IT support as well as the computers themselves – is traditionally a cost-prohibitive barrier. Second, technology is always changing. If a student enters a four-year university to study a field as dynamic as analytics, the information they learn as a freshman may be somewhat outdated by the time they graduate.

How do you start? It starts by helping students learn the fundamentals through the most advanced technologies. With this experience, they can be prepared to take on the challenges of the future via analytics.

What Today’s Students Need to Learn
While terms like “big data,” “deep learning,” “cognitive computing” and “machine learning” garner all the buzz, analytics – which serves as the encapsulation of all of these buzz words – is like any other field of study. Students first have to understand the basics. After all, you don’t start calculus without a solid understanding of algebra and trigonometry. The same applies to analytics: Students have to understand some of the fundamental aspects before hitting the job market.

There are a variety of concepts that any analytics professional needs to know. Here are three areas that can provide a foundation for a long career in analytics:

Descriptive analytics. Descriptive analytics (especially in statistical terms) have been around for hundreds of years. The first well-publicized population data collection project took place in 1749 in Sweden. The Swedish government wanted to know the spread of its population to help allot military resources more effectively. While the project was rather simple (by today’s standards), all of the statistics and mathematics were manual, making this quite an undertaking for an 18th-century organization.

Put simply, descriptive analytics provides a summary of collected data points using a sample of the population or an entire data set. A descriptive analytics model help you understand what happened and to some extend why it happened. Although it is the foundation of other forms of analytics, there are plenty of metrics of this kind still in use, including:

central tendency (mean, median and mode)
dispersion (variance, standard deviation, range)
relationship (covariance, correlation, cross-tabulation)
augmented with simple graphics (histogram, box-plot, scatter plot, correlation, regression)
business intelligence
on-demand/ad-hoc reporting
multi-dimensional modeling / OLAP (cubes, drill-down/roll-up, slicing/dicing)
data warehousing (to support decision support)
KPI visualizations (dashboards and scorecards)
A good understanding of these basic elements business reporting is fundamental, for everything from a broader career in data science to more specialized knowledge in things like cognitive computing and machine learning.

Predictive analytics. Predictive analytics builds on descriptive statistics and uses more advanced statistical algorithms and machine-learning methods. The goal is to identify likely future outcomes based on historical data. With predictive analytics, users can go beyond a snapshot of the way things are. With a better forecast, predictive analytics can deliver new insights that lead to better, more effective actions.

Predictive models use previous results to develop models that can be used to predict values for different or new data. The modeling results in predictions that represent a probability of the target variable (for example, future revenue) based on a set of input variables. This is different from descriptive models that help you understand what happened or diagnostic models that help you understand key relationships.

A 2014 TDWI report found that the top five things predictive analytics are used for is to: 1) identify trends, 2) understand customers, 3) improve business performance, 4) drive strategic decision-making and 5) predict behavior.

Prescriptive analytics. Prescriptive analytics is the highest echelon in the three-layered analytics hierarchy. The goal is to identify the best (i.e., optimal) decision (i.e., choice) from the large (sometimes infinite) number of feasible options. While descriptive and predictive analytics establish the landscape of options and choices, prescriptive analytics is used to sort them out and find the most plausible one to base the decision on. The techniques used in prescriptive analytics include optimization (linear, non-linear), simulation (stochastic and deterministic enabling a wide range of what-if analyses), and multi-criteria decision-making techniques.

Figure 1 depicts the three progressive (in terms of the degree of analytics sophistication/intelligence) layers of analytics along with the questions answered and techniques used for each of them.


\
Figure 1: Progressive layers of analytics (descriptive, predictive and prescriptive). Source: Delen, D., 2015, “Real-World Data Mining: Applied Business Analytics and Decision Making,” FT Press.

Naturally, this is a field that continues to grow and develop as new techniques emerge and technologies adapt to use these new techniques. Students studying predictive analytics will have the ability to go into some of the more cutting-edge fields of study in the field of analytics.

A critical use of prescriptive analytics comes with data that has not been traditionally part of many analytics environments. With the advent of big data, the scope of analytics has widened to include processing of what is often called unstructured or semi-structured data (implying that the data is not necessarily structured for the computer programs to readily process). This would include things like text, audio, images and video.

For decades, analytics focused on tabular, database-driven data. This information was easier to organize and orient for analysis, especially when compared with a text document. While there is metadata within a table to explain what’s in each column or row, there are few such indicators in a text document.

However, text files are a potential gold mine of insights that could help an organization better understand its overall health and day-to-day operations. Users in today’s analytics environment need to understand how to use unstructured and semi-structured text such as invoices, comments and other free-form fields.

Text analytics is becoming more important in the era of social media. One of the goals of text analytics is to accurately extract entities, facts and, more importantly, sentiment. The information from feeds from Twitter, Facebook and other social media networks can provide information about a company’s perception among its customers and targeted customers. If students have the ability to manipulate and review this type of information, their skills will be in high demand.

Tools to Educate Next Generation of Analytics Professionals
Understanding the philosophies and techniques behind analytics is only part of the equation. While in school, students need hands-on training to apply these techniques to learn the strategies needed to succeed in the business world. There are a host of technologies available that can help students start developing the skills they will need in the professional world.

Many universities around the United States and abroad have started degree programs specific to analytics both at the graduate and undergraduate level. Having a degree in analytics or a closely related field has become a significant differentiator in the job market. Coupled with hands-on experience and technical skills, analytics students often receive multiple attractive job offers even before they graduate.

Putting analytics in the hands of students is now easier than ever. Once, professors needed labs full of expensive desktops and servers – along with purchased software – to teach the latest techniques in analytics. This was a cost barrier that many schools could not surmount.

Today’s academic environments can employ much more cost-effective and nimble environments to create a responsive training environment in analytics. Technology advances have improved both the delivery of software and their power with the uninitiated user. All of this is helping more schools put the power of analytics into the hands of their students.

Cloud-based deployments. One of the most powerful changes in the software world has been the delivery of technology via cloud-based technologies. By having software deployed through a massive, virtual network of computers, universities can deploy and maintain analytics technology more easily than ever.

The key benefit comes on the hardware side. Accessing cloud-based applications requires nothing more than an Internet browser. Some even work on a mobile device. So, a university setting up a high-end analytics lab doesn’t need to purchase a set of powerful desktops and servers. A simple laptop connected to the Internet can provide access to the latest technologies.

Cloud-based capabilities make it easier for software providers to get technology in the hands of students. An example would be the Teradata University Network (TUN) [1], a consortium of companies working together to deliver analytics to college students. TUN provides tools and resources to almost 5,000 registered faculty members from 2,500 universities in 115 countries to ensure their students receive experience with various technologies. This would be difficult with a traditional model of software deployment, but with cloud deployments, it’s easier to put the right technology in students’ hands.

TUN is not limited to just cloud-based software platforms; it also contains a variety of teaching resources from syllabus material to case studies, data sets and scripted videos. Starting to use TUN as an analytics instructor requires only three steps (see Figure 2).


Figure 2: Three steps to using TUN.

Easy to use technology. For years, the use of analytics technology required a deep understanding of a coding language to manipulate the data for analysis and perform the analytics tasks. Because of this requirement, the initial generations of analytics professionals were, in some ways, more like application developers than analysts. They spent much of their time building and refining models, and the outcome went to a business person to evaluate and use to make decisions.

A modern analytics environment operates in a much different way. Today’s analysts – sometimes called data scientists – use GUI-based tools to replicate the work that was once done at the code level. They can visually make connections in the data and try many different techniques to see how a new analytics technique works.

In the academic world, it’s important that the next generation of analytics professionals is aware of these more modern interfaces and how to apply them to real-world challenges. SAS Visual Analytics, for example, is a technology that allows users to visually explore data and see how to make connections between different data elements. Figure 3 illustrates a dashboard that shows relationships in different data elements through several charts and graphs.


Figure 3: Screen shot of SAS Visual Analytics.

This type of technology is becoming more commonplace in the business world. It allows people who are doing the analysis to work more closely with their business counterparts. Students can use this technology to understand how to collaborate on analytics projects in school, and then take those skills to the business world. For instance, MBAs, less technically savvy university students – the business managers of the future – can easily use these tools to understand and appreciate the value of data and analytics in the world of business management.

Dursun Delen (dursun.delen@okstate.edu) is a professor of business analytics within the Department of Management Science and Information Systems at the Spears School of Business, Oklahoma State University.

Leave a Reply

Your email address will not be published. Required fields are marked *