Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

nabqr: Python package for improving probabilistic forecasts

Bastian Schmidt Jørgensena, Jan Kloppenborg Møller, Peter Nystrup, Henrik Madsen

TL;DR

Uncertainty quantification in ensemble forecasts is critical for reliable decision-making in renewable energy and related sectors. The authors introduce NABQR, a Python package that first corrects ensemble forecasts using LSTM networks and then applies time-adaptive quantile regression to produce calibrated probabilistic distributions. The work provides a modular, open-source implementation with an end-to-end pipeline, multiple scoring metrics, and data-simulation utilities. Empirical results for wind power in Denmark show up to 40% improvements in mean absolute error, highlighting practical benefits for grid operations and renewable integration.

Abstract

We introduce the open-source Python package NABQR: Neural Adaptive Basis for (time-adaptive) Quantile Regression that provides reliable probabilistic forecasts. NABQR corrects ensembles (scenarios) with LSTM networks and then applies time-adaptive quantile regression to the corrected ensembles to obtain improved and more reliable forecasts. With the suggested package, accuracy improvements of up to 40% in mean absolute terms can be achieved in day-ahead forecasting of onshore and offshore wind power production in Denmark.

nabqr: Python package for improving probabilistic forecasts

TL;DR

Uncertainty quantification in ensemble forecasts is critical for reliable decision-making in renewable energy and related sectors. The authors introduce NABQR, a Python package that first corrects ensemble forecasts using LSTM networks and then applies time-adaptive quantile regression to produce calibrated probabilistic distributions. The work provides a modular, open-source implementation with an end-to-end pipeline, multiple scoring metrics, and data-simulation utilities. Empirical results for wind power in Denmark show up to 40% improvements in mean absolute error, highlighting practical benefits for grid operations and renewable integration.

Abstract

We introduce the open-source Python package NABQR: Neural Adaptive Basis for (time-adaptive) Quantile Regression that provides reliable probabilistic forecasts. NABQR corrects ensembles (scenarios) with LSTM networks and then applies time-adaptive quantile regression to the corrected ensembles to obtain improved and more reliable forecasts. With the suggested package, accuracy improvements of up to 40% in mean absolute terms can be achieved in day-ahead forecasting of onshore and offshore wind power production in Denmark.

Paper Structure

This paper contains 9 sections, 3 figures, 2 tables.

Table of Contents

  1. Motivation and significance
  2. Software description
  3. Software architecture
  4. Software functionalities
  5. Submodules
  6. Illustrative examples
  7. Impact
  8. Conclusions
  9. Example of Simulated Data

Figures (3)

  • Figure 1: Overview of pipeline in NABQR. Note that for the first step if data is not specified to be loaded the pipeline will automatically simulate data.
  • Figure 2: Excerpt of TAQR predictions with prediction interval 5%-95% shown in shades of blue for Denmark (area DK2) Offshore wind power production. Obtained by running run_nabqr_pipeline on wind power production.
  • Figure 3: Simulation for 250 hours of normalized wind power data including ensembles with the method simulate_wind_power_sde in run_nabqr_pipeline.