We asked five social influencers how data scientists can collaborate to build better business applications rapidly. See what they had to say in response about collaboration best practices, tools and more.
Bob E. Hayes
“They need to be provided a process or platform that improves how different data scientists collaborate with each other. In fact, research shows that different aspects of teamwork quality impact the success of projects involving innovation. This process/platform needs to improve team communication, coordination, balance of member contribution, member support, effort and cohesion. Toward that end, I recommend data science team members review the scientific method to organize their collective efforts to extract insights from data. Additionally, while many vendors provide cloud data services for each step of the analytics life cycle (idea, data collection, data management, analytics, reporting), the ones who will win will better integrate the analytics journey to facilitate teamwork.”
Bob E. Hayes is chief research officer at AnalyticsWeek and president of Business Over Broadway. Hayes conducts research in the area of big data, data science and customer feedback—for example, identifying best practices in customer experience programs, reporting methods and loyalty measurement. And he provides consultation services to companies for helping them improve how they use their customer data through proper integration and analysis.
“The relationship and collaboration between data science and data engineering is incredibly important to building real products faster—particularly what those each mean for the team or company that you’re working within and the goals of each group. Well-defined lines between these two teams keeps you all building in the right direction and enables data engineering to plan and invest in the right infrastructure to ensure that you can build what’s needed to meet the business problems. Additionally, ensuring good product and business input to data science problem definition is key. Technically, ensuring consistency in standards and approach allows you to make best use of the infrastructure set up and be able to take models from test to production quickly.”
Chris Maddern is cofounder of Button, a leading-edge marketplace for app connections. Prior to Button, Maddern led mobile engineering at popular social payments network Venmo, and has founded several mobile products startups—some of which were successful.
“There needs to be closer interaction with the business, both formally and informally. Too often, data science teams work in isolation from the mainstream business. Business users need to be able to share in the tools, at some basic level, to create formulations to their own queries.”
Joe McKendrick is an author, independent researcher and speaker exploring innovation, information technology trends and markets. McKendrick’s Forbes.com column, “Disruptions,” explores how technology innovations move markets and careers.
“We are keenly aware of the analytic product challenge of bridging the gap between development and deployment. Collaboration between data science and software engineering teams is vital to overcoming this. Ideally, this collaboration occurs in the same physical space, using common tools (for instance, white boards) and within the problem environment. This ensures that the full knowledge and experience of the team can be employed to fail fast and ultimately create innovative solutions.
Unfortunately, this synergy can be hard to recreate outside of a business unit focused on common problems. At the same time, the reading and writing of blogs only gets you so far to this goal. So, it is essential for data scientists to actively collaborate at venues with broader perspectives. This can be at large conferences such as the Predictive Analytics World, Spark Summit, Strata + Hadoop, the ACM’s Conference on Knowledge Discovery and Data Mining, or the INFORMS analytics conference. It could also be at more localized events such as meet-ups focused on general data science or specific topics within the industry.”
Dave Saranchak is a data scientist with Elder Research, where he develops and applies statistical data modeling techniques for national security clients. Saranchak developed and leads training for the Elder Research office in Baltimore, Maryland, emphasizing the technologies best able to meet clients’ needs.
“Rapid development can be a challenge for data scientists, especially anyone who has limited experience and exposure to product development. Unlike programmers who can defer to technical requirements, and business leaders who use goals and objectives, the breadth and depth of a data scientist’s skill set can make finding the right fit within a project challenging for product managers.
Data scientists should always be looking for good data science opportunities when working within a team. Successful collaboration requires understanding the ins and outs of the business—or organization—well enough to find where the data can fit into the current process and then translating this opportunity into technical specifications for developers and engineers.”
Jennifer Shin is founder of and principal data scientist at 8 Path Solutions, a data science, analytics and technology company. Shin is a recognized thought leader, data science contributor for the IBM Big Data & Analytics Hub and technology expert on eHow.com.
Other data science opportunities
To build high-quality data science applications rapidly, try IBM’s new Data Science Experience for yourself. Also, don’t forget to explore how Spark, R and open data science can help you build your own apps in less time than ever before.
Source: IBM Big Data