A Cognitive Load Perspective on the Design of Blocks Languages for Data Science

The difficulty of learning data science is believed to arise from its deep prerequisites in statistics, programming, and machine learning. This poster explores how blocks languages may be used to reduce the cognitive load of learning data science. Unlike blocks languages for introductory programming, blocks languages for data science naturally align with a high level of abstraction and almost exclusively sequential execution. However, these gains in simplicity are offset by the high level of parameterization at the block level. Three designs for blocks languages are presented and compared, and implications for abstraction, sequential execution, and parameterization on cognitive load are discussed.