Does pair programming work in a data science context? An initial case study

While pair programming has been studied extensively for software programmers, very little has been reported with respect to pair programming in a data science project. This paper reports on a case study evaluating the effectiveness of pair programming within a data science / big data context. Our findings show that pair programming can be useful for data science teams. In addition, while the driver role was similar to what has been described for software programmers, we note that the observer role had an expanded set of responsibilities, which we termed researcher activities. Further exploration is required to explore if these expanded roles are specific to data science pair programming.