As machine learning grows in prominence, adoption in high-impact use cases such as anti-fraud and network security are growing. Having a high performing statistical model in these areas are critical: a false positive error leads to unnecessary work, while a false negative error increases exposure to potential threats. Since there are no perfect machine learning models, as data scientists our task is to first convince ourselves and then convince others that we have a statistical model worthy for production. Persuasion, though, can be difficult because many of the steps and assumptions that go into training a statistical model from data are difficult, if not impossible, to accurately share with the ultimate consumers of the model.
Drawing on ideas from the philosophy of science such as falsifiability and counterfactuals, we present a framework for triangulating the performance of machine learning models using a series of questions to help establish the validity of performance claims. In navigation tasks, triangulation can be used to determine one’s current location based on the angle and distance from other landmarks with known position. We believe triangulation of a different sort is necessary to determine the performance of machine learning models. Each of the steps that go into making a machine learning model including input data selection, sampling, outcome variable selection, feature creation, model selection and evaluation criteria shape the final model and provide necessary context for interpreting the performance results. Our framework highlights ways to uncover assumptions hidden in those choices and identify higher performing models.
You need this ticket from Eventbrite to sign up:
Applied Machine Learning Conference.