This message was deleted.

I have not been able to find the answer in searching github pages (e.g., issues, discussions, or release notes) nor slack channels.

Hi @Brian Head! Very interesting question. Conformal prediction si currently only available in Stats and ML. We have plans to add it to NF in the future. All NF models have the option to produce probabilistic outputs simply by adjusting the training loss. For example, with the

MQLoss

to output multiple quantiles, or with the

DistributionLoss

class that assumes some underlying distribution. Both type of losses are viable options that will produce calibrated PI in most cases. This is the standard approach for deep-learning methods (for eg. refer to the DeepAR paper). Here are some tutorials: https://nixtla.github.io/neuralforecast/examples/uncertaintyintervals.html https://nixtla.github.io/neuralforecast/examples/longhorizon_probabilistic.html