Statistics seminar: Ryan Lekivetz

Ryan Lekivetz
JMP

Room: M3 3127

Applying Design of Experiments to Statistical Software Testing

Testing statistical software presents unique challenges: developers often double as testers, formal training in software QA may be limited, and time constraints are common. This talk explores how principles from Design of Experiments (DOE) can be leveraged to create efficient and effective test strategies for statistical software. By treating test case construction as a designed experiment, we can uncover subtle bugs, improve test case coverage, and enhance confidence in results. The discussion also highlights practical considerations and lesser-known testing techniques tailored to the realities of statistical software development, illustrating how statistical thinking can elevate software quality assurance.