Copulas for uncertainty quantification in time series forecasting

On Friday (2024-07-26), I presented nnetsauce (“Probabilistic Forecasting with nnetsauce (using Density Estimation, Bayesian inference, Conformal prediction and Vine copulas)”) version 0.23.0 at an sktime (a unified interface for machine learning with time series) meetup. The news for 0.23.0 are:

• A method cross_val_score: time series cross-validation for classes MTS and DeepMTS, with fixed and increasing window
• Copula simulation (thanks to pyvinecopulib) for uncertainty quantification in classes MTS and DeepMTS:
  • type_pi based on copulas of in-sample residuals: vine-tll (default), vine-bb1, vine-bb6, vine-bb7, vine-bb8, vine-clayton, vine-frank, vine-gaussian, vine-gumbel, vine-indep, vine-joe, vine-student
  • type_pi based on sequential split conformal prediction (scp) + vine copula based on calibrated residuals: scp-vine-tll, scp-vine-bb1, scp-vine-bb6, scp-vine-bb7, scp-vine-bb8, scp-vine-clayton, scp-vine-frank, scp-vine-gaussian, scp-vine-gumbel, scp-vine-indep, scp-vine-joe, scp-vine-student
  • type_pi based on sequential split conformal prediction (scp2) + vine copula based on standardized calibrated residuals: scp2-vine-tll, scp2-vine-bb1, scp2-vine-bb6, scp2-vine-bb7, scp2-vine-bb8, scp2-vine-clayton, scp2-vine-frank, scp2-vine-gaussian, scp2-vine-gumbel, scp2-vine-indep, scp2-vine-joe, scp2-vine-student

For more details and examples of use, you can read these slides:

https://www.researchgate.net/publication/382589729_Probabilistic_Forecasting_with_nnetsauce_using_Density_Estimation_Bayesian_inference_Conformal_prediction_and_Vine_copulas