Good data scientists are hard to find! This is in spite of the rapid growth of University master’s degree programs and commercial boot camps. Those sources do produce a much-needed supply for the data scientist demand projected over the next decade. However, becoming a good data scientist takes more than what these programs can provide.
Data Science is unique in having roots in multiple professions -- including statistics, programming, business analysis, communication, and strategy. We are mutts as opposed to pure breeds; we can’t focus on just one discipline but need to be great in many. The most advanced can envision, plan, and execute not only one piece of the puzzle but the complete end-to-end project. That takes training and experience, and to succeed with analytics, organizations must invest in talent for the long term and keep turnover low within the analytics team. To achieve that, focus on the following key guidelines.
Being able to attract skilled data scientists that fit your company culture is obviously a basic requirement, but it is also essential to nurture a collaborative team-based environment conducive to learning rather than a competitive one. For instance, Peer reviews and having teams compete on the same or similar tasks can create a competitive environment. You want a place where all ideas and implementations can be challenged, and team members are encouraged to learn from each other. Remember, data scientists like to work on data science projects. When they are assigned to reporting or data engineering problems, it is a mismatch of their skill set and interest. If they are mis-assigned too long, not only will the organization not fully benefit from their talent, but they will soon look elsewhere for the challenge they find motivating.
Allow Them to Learn and Grow
Allowing enough time and opportunities for data scientists to develop skills and experience is expensive. It can be painful from the business perspective as the return on investment is not obvious or immediate. But if they stay with you, it can be enormous. You must get them out of their comfort zone and exposed to new ideas. At Elder Research, we use several different approaches:
This tradition dates to early days of Elder Research where an internal or an external person or team gives a weekly 45-minute company-wide presentation about their project or research. These talks are mostly technical, but we do include executive-level talks as well. Inviting guest speakers provides an outsider’s perspective or information about a new technique that we have not used. Talks are followed or preceded by a catered lunch, with outside guests invited, to foster teamwork and discussions.
Scientific Paper Book Club
This gets our technical folks really excited and is best done locally at each office. Each week the team reviews and discusses a scientific paper related to Data Science, Machine Learning, Statistics, Software Development, or Technology. The paper selected could be a landmark paper cited by thousands of researchers or a new state-of-the-art method that recently generated a lot of buzz. Data scientists are immersed in science and research, but the field is still young, so even though most of us fall into the category of “practitioner” rather than “researcher” it is important to continuously learn and improve our skills. As time progresses your team will gain understanding of the most commonly used approaches and will gradually develop the skills required to implement more advanced techniques. There are several methods for running a reading club, but at Elder Research we assign one person each week to select, read, and present the idea behind a paper to the group. The resulting discussion creates a lot of excitement around the science and is an excellent team-building opportunity.
Massive Open Online Courses (MOOCs)
MOOCs can be a great learning platform for continuing education. They are cost effective and data scientists can take them in-between projects or on their own time. Data Science related courses cater to different levels of experience and skills, from beginner to advanced. MOOCs allow data scientists who are keen to learn new skills an efficient way to stay on the cutting edge and bring greater value to their team.
Elder Research encourages on-line continuing education, such as Coursera, EdX, Udacity, etc. Individuals who sign up for the services are expected to finish at least four courses or a specialization each year. We also have done courses as a team, which facilitates creative discussion and helps overcome road blocks.
Data science and related conferences provide great opportunities for your team to listen to speakers present their work, exchange ideas, and to network. Sending team members to key conferences allows them to meet top practitioners and learn about the recent developments in the field. They can then share their experiences with the team when they return to the office. And, we encourage our colleagues to present their work, in conjunction with our clients, at conferences, where they can share their breakthrough ideas with the larger community.
Workshops and In-person Training
In-person workshops are ideal for more advanced and specialized technologies for which MOOCs are not available, online documentation is not sufficient, or a certification exam is required. At Elder Research, we send individuals to workshops who are keen to gain specific skills for which there is a strategic business interest. We are also a leading teacher of analytics workshops in the U.S. and occasionally overseas. And, the best way to learn a subject is to have to teach it! (There is nothing like the specter of public humiliation to motivate you to get your thoughts in order.) As with presenting a paper in the reading club, or presenting original work at a conference, the exercise of curating material to distill the most important points to communicate them well, is a fantastic way to master the material.
To summarize, sustaining a team culture of learning, in action, is vital to attracting, developing, and retaining the high-quality data scientists that will bring great value to your organization.
Download the eBook Top 10 Data Mining Mistakes to learn about other mistakes to avoid in your analytics projects.
Read the blog Building an Effective Data Science Team
Read the blog Obstacles and Opportunities for Women in Data Science