JSoC 2015 Project: Development of NullableArrays.jl for Julia

This article details the development of the NullableArrays.jl package, a project completed under the 2015 Julia Summer of Code program. The author, a PhD student in statistics, created the NullableArray data type to efficiently handle missing values in the Julia programming language. Unlike the existing DataArray structure, which suffered from significant performance penalties due to type instability and excessive memory allocation, NullableArrays utilize a single Nullable{T} type. This design allows Julia’s compiler to perform effective static analysis and generate efficient machine code. Benchmark tests demonstrated that operations on NullableArrays are comparable in speed and memory usage to standard arrays, offering a substantial improvement over previous methods. The project was supported by MIT, NumFocus, and the Gordon & Betty Moore Foundation, with mentorship from John Myles White. Much of the work was conducted at the Recurse Center. The article highlights the technical advantages of the new package and acknowledges the contributions of the open-source Julia community to the author's growth as a programmer.