I recently caught up with Vinod Bakthavachalam, Data Scientist at Coursera, to discuss how to build in-demand skills in data science such as machine learning, statistics, and data management across your organization to drive competitive advantage. Vinod is a data scientist working with the Content Strategy and Enterprise teams where his work has recently focused on developing ways to measure the learning outcomes from taking Coursera classes, especially in the context of company sponsored training. Prior to Coursera, he got triple Bachelors degrees from UC Berkeley in Economics, Statistics, and Molecular and Cell Biology, and his Masters degree in Statistics from Stanford.
insideAI News: Please give our readers a high-level view of Coursera for Business and how you upskill employees to drive competency in high demand fields like data science.
Vinod Bakthavachalam: Coursera for Business provides companies with the world’s best learning experience and best content to transform their talent. We help equip workforces with the knowledge to compete today across both hard and soft skills in subjects like data science, business, and computer science. Our courses come from over 170 top universities and companies around the world, and we work with our customers to curate learning programs that align to their corporate strategy and learning needs.
insideAI News: When speaking with a new potential customer, how do you assess what data-related skills the company may need to develop and help guide their learning?
Vinod Bakthavachalam: We first ask some fundamental questions about their analytical capabilities across the organization, to understand their use case and establish the right learning strategy. These questions are:
- What is data? Do they have a shared high-level understanding and the appropriate infrastructure in the company to power downstream analysis and insights? This relates to data management expertise.
- How do you analyze data and derive business value as well as scale these insights across the organization and to customers? This is where data analysts (Excel, SQL) and data scientists fit in.
- How do you communicate the insights and products created from data? This is where data visualization/communication comes in.
Once we have this information, we customize a learning program to fit a customer’s specific needs.
insideAI News: Here is a question I get asked all the time — what would you advise for an aspiring data scientist? What programs are available?
Vinod Bakthavachalam: Data Scientists perform sophisticated empirical analysis to understand and make predictions about complex systems. When we look at the skills of data scientists on the Coursera platform and for our own hiring, we look at ability across a key set of competencies that are essential to the role:
- Math (calculus, linear algebra)
- Statistics (linear regression, hypothesis testing)
- Programming (R, Python)
- Data management (assembling and cleaning data, SQL)
- Machine learning (supervised and unsupervised learning)
- Data visualization/communication (discussing and presenting results)
Data scientists should have a t-shaped skill profile whereby they have a strong foundation in each of those areas, but then specialize based on the type of work they want to do. For example, all data scientists need to know things like linear algebra, AB testing, and SQL. But data scientists who want to do more inference/experimentation need a stronger skill set in statistics while those wanting to do data products such as recommendation systems need more machine learning.
At Coursera we designed our data science catalog to teach both foundational and advanced skills in the above competencies. This way companies can train an entire analytics organization on Coursera and allow their technical talent to specialize in the diverse areas that are required to fully capture the value that data can provide. Content could include Statistics with R and Deep Learning.
Aspiring data scientists then should first acquire the broad skill set across key competencies and then specialize in a particular area by taking more advanced content, guided by the type of work they want to do.
It is also important to practice applying these skills. Data itself is not valuable and it is up to data scientists to create business value using it. Our courses place an emphasis on applied projects to give learners rigorous opportunities for practice.
insideAI News: What would you say to employees with little to no quantitative background who are increasingly expected to collect, analyze and communicate data to make business decisions?
Vinod Bakthavachalam: Employees who have little to no quantitative background will need to do three things.
First, they should work to acquire a basic data savviness, which will require some foundational math and data skills. As the amount of data grows and its importance within businesses increases, more employees will be required to incorporate data into their decision making, so having the theoretical knowledge to do basic analysis and understand the language of data will be critical.
Second, people will need to be able to think in a data first way. The technical knowledge is necessary but not sufficient. Similar to how data scientists need to practice applying their skills, employees who need to make data driven decisions will need to practice applying their data skills in order to build this muscle.
On Coursera we have courses and specializations specifically designed for these functional analysts and other decision makers who need to apply data skills in a specific domain. The courses are less technical than those for data scientists but still teach core concepts in an applied manner through business focused case studies. Content could include Data Analysis Using Excel and Business Metrics for Data Driven Companies.
Finally, businesses will need to create operational processes to ensure that these functional analysts and other decision makers can work seamlessly with data scientists. As data driven decisions become more of a business first principle, ensuring people can communicate effectively with data scientists and leverage the value of their work will be critical in having a consistent and unified data strategy across the organization.
insideAI News: What about data scientists and software engineers with some coding and linear algebra experience? What can you offer this class of employee?
Vinod Bakthavachalam: Individuals who know coding and some linear algebra likely have the broad foundational skill set in programming and math to succeed in data science. It is then about rounding out the foundation in the other competencies (such as statistics, machine learning, etc.) and identifying what area to specialize in given interests and the desired types of projects.
On Coursera we can tailor curriculums by combining various courses and specializations that do both of these things through the creation of learning paths for individual employees based on their objectives and background. Content could include Machine Learning with TensorFlow and Big Data for Data Engineers.
We are also developing career pages and diagnostics to help employees figure out the type of data science work they might prefer based on their interests and prior background.
insideAI News: Lastly, how about the non-practitioner class of employee, maybe senior level people who need a foundation in statistical and data science techniques in order to hire, interpret, and work with data scientist and business analysts?
Vinod Bakthavachalam: For senior level people the emphasis should be on ensuring that everyone in their team or company has the internal infrastructure, requisite skill set, and operational processes to be a data first company. This will involve investing in building the right infrastructure to ensure people can access the data they need and have the right tools to analyze it, training their employees in the needed skills to empower data analysis and conversations around it, and ensuring that data analysis is built into the company’s execution plan.
On Coursera we have courses designed for these senior level people that provide a macro view of how to do the above things. We also have internal experts that can design a curriculum targeted to different roles within a company to ensure that the entire organization from the non-technical to technical talent has the skills they need to execute on a data first strategy.
Lastly, senior level people should lead by example and ensure they themselves have an understanding of what data is and how to use it. We have courses to help them see why data is important and how to learn enough to guide their employees here. Content could include Executive Data Science and What is Data Science.
Find out how you can build data skills across your organization. Learn more.
