- 16th April 2018
- Posted by: Manolis
If you are a regular reader of this blog, you are aware of my emphasis on data. Today’s post focuses on data scientists: our need for them, how we should train them, and the role that entrepreneurship in the education space has to play (and is playing) in this endeavor.
P-20 education is increasingly emphasizing data fluency — the ability to understand the presentation of data, to infer meaning, and to visualize data. Organizations like Digital Promise, for instance, now offer micro-credentials in data fluency (among many topics) for educators. Indeed, my Innovation @ Penn GSE group will offer a computational thinking certificate for educators beginning in January 2018. An introductory data science class at UC Berkeley shows that while data fluency is essential, data scientists must go further: forming questions; collecting and cleaning data; conducting exploratory analysis, visualization, statistical inference and prediction; using programming languages and algorithms; and engaging in informed decision-making. Data scientists analyze and interpret complex digital data, particularly the “big data” constantly emerging from our lives in the digital world, and use these insights to drive business decisions.
Much ink has been spilled arguing for more STEM in schools and greater data fluency — we don’t need to go over this ground. More interesting, from my point of view as an educator, investor, cultivator of startups and general champion of edtech, is the role education entrepreneurs and the startup community can and should play in developing and deploying data scientist programs and training.
Who Is Training the Data Scientists?
Many entities are engaged in this work already: General Assembly, Data Society, and Springboard are examples of for-profit education companies that provide data scientist training in blended educational settings. Fully-online Western Governors University and traditional, perennial ranking kings such as Caltech, Berkeley (etc.) already offer high quality degrees in data science. The problems with and benefits of these types of offerings are well chronicled, and include: opportunity costs, significant financial outlay, and the problem of — in many cases — having to complete an entire degree. This is not to mention the problematic assumption that everyone has equal access to high-speed internet, as well as the basic mathematical competency to undertake these courses.
MOOC-providers address these issues to some extent: Udemy, for instance, offers a low-priced, self-paced course on data science (which one seriously in-depth review of data science programs has dubbed “the best”). Luckily, foundational courses — such as statistics and certain programming languages — can also be had for a pretty low fee.
Other edtech companies fill the need for employers to be able to verify that particular skills have been developed, for instance through the use of non-traditional mediums like MOOCs. One company in this space, Degreed, uses an assessment process to certify competency in specific skills. The company is agnostic when it comes to how or where those skills were acquired, an approach that can help level the playing field for individuals.
Numerous players are working to spark interest in the skills that undergird data science. Some focus on underserved groups, such as Philly-based nonprofits Coded by Kids and TechGirlz. Others are extracurricular enhancements, such as the (nonprofit) FIRST Robotics program. TROBO, Pivot Interactives, and Mosa Mack Science are just a few of the STEM-focused companies we’ve had in the Milken-Penn GSE Education Business Plan Competition in the last few years. But there are many more people who would benefit from such training, and yet require serious development of the intermediary skills.
What Role Should EdTech and Education Entrepreneurship Play?
DataKind is a nonprofit that manages a network of volunteers — previously-trained data scientists — to glean insight from nonprofits’ data, and thereby to drive decision-making. DataKind solves organizations’ problem of not having enough funding to have their own data scientists. Yet it still does not solve the underlying problems of nonprofits not having enough money to fund a data position, and of not having enough well-trained people to fill such roles. To break the cycle, we must take collective action:
• Encouraging nonprofits and other players to think creatively about impact and impact measurement
• Encouraging nonprofits and early stage companies to hire data analysts and scientists
• Encouraging stakeholders to hold ventures and organizations to high standards of data collection, analysis, and dissemination of findings
• Calling on foundations and grantmakers to specifically fund data analyst and data scientist positions at nonprofits
• Demanding earlier, more robust opportunities for students to develop data fluency, STEM experience and interest
We need to go further than DataKind. It is time to develop a social impact company (or network of companies) that finds, trains, and manages data analysts who could eventually become data scientists — a sort of bridging organization. As apprentice data scientists, trainees could work with 5-10 nonprofits at a time on those nonprofits’ needs. Such a company would create pathways to employment for individuals who may have traditionally been under-qualified for such jobs, and would simultaneously help meet the growing demand for useful impact data to come from nonprofits. It would also produce well-trained individuals who could fill data scientist roles at nonprofits as well as early stage companies that can’t afford a full-time data scientist. Alternatively, this could be structured as an AmeriCorps type program, wherein a couple of years of service pays for the cost of educating that person (either by paying back loans or earning dollars for use in accredited programs).
Is there room in a crowded landscape of pay-to-play credentialing (this includes traditional higher education) for social impact companies that can tackle equity gaps in who gets to be a data scientist (and, by extension, whose families, companies and communities benefit)? I think so: data scientists are in demand — and paid well — as big data and its uses and applications increasingly drive business. Big data is now a fact of life across all sectors of our economy. That makes training individuals in the skills related to data (and its uses) a tool for economic inclusion and greater equity, and a rising tide that lifts all boats.