Sampling Bias Correction for Supervised Machine Learning: A Bayesian Inference Approach with Practical Applications

Given a supervised machine learning problem where the training set has been\nsubject to a known sampling bias, how can a model be trained to fit the\noriginal dataset? We achieve this through the Bayesian inference framework by\naltering the posterior distribution to account for the sampling function. We\nthen apply this solution to binary logistic regression, and discuss scenarios\nwhere a dataset might be subject to intentional sample bias such as label\nimbalance. This technique is widely applicable for statistical inference on big\ndata, from the medical sciences to image recognition to marketing. Familiarity\nwith it will give the practitioner tools to improve their inference pipeline\nfrom data collection to model selection.\n