This article was accepted into the Tunable ML workshop at NeurIPS 2023.
Conformal prediction (CP) is a method of estimating risk or uncertainty when using machine learning to help comply with common risk management regulations often seen in fields such as healthcare and finance. CP for regression can be challenging, especially when the product distribution is heteroscedastic, multimodal, or skewed. Some of the issues can be addressed by estimating a distribution over production, but in reality, these approaches can be sensitive to estimation error and produce unstable intervals. Here, we avoid the challenges by converting regression into a classification problem and then use CP for classification to obtain sets of CP for regression. To preserve the order of the continuous output space, we design a new loss function and introduce necessary modifications to CP classification techniques. Empirical results from many benchmarks show that this simple approach gives surprisingly good results on many practical problems.
