stat This key contains the final assessment metrics, organized into subkeys. Each subkey has a Pandas DataFrame, with the forecasting methods on the rows and the forecasting horizons on the columns. statSeries This key contains the time series of the Continuos Ranked Probability Score (CRPS) and its decompositions (i.e., uncertainty, reliability and resolution). The subkeys are organized as follow: Forecasting method Forecasting horizon (in hours ahead) The assessment metric: Pandas DataFrame with time series. Note that the DateTime is related to the NWP base time. evalSeries This key includes the forecasted time series. Each forecasting method has a corresponding subkey. Additionally, there is a subkey containing the observed time series shifted to align with the forecasts (referred to as 'target'), and a subkey for the average interquartile range of the weather-to-power model (referred to as pcm_iqr). The forecasting methods subkey is organized as follow: Forecasting horizon (in hours ahead) Output type: quantile, CDF and PDF reliability This key includes the data used to plot the reliability diagrams. Each forecasting method has a corresponding subkey, which contains two Pandas DataFrames. The 'value' DataFrame contains the observed frequencies, and the 'confBar' contains the consistency bar related to each quantile (estimated using a binomial distribution).

This dataset contains files with the predicted time series and the forecast assessment conducted in the paper Seamless short- to mid-term probabilistic wind power forecasting. It is licensed under the Creative Commons Attribution 4.0 International (CC BY). The files are organized into subfolders named according to the case study. The case studies are called according to the year used as the test period and the window length used in the estimation period. For instance, 2023-4y refers to 2023 as the test period and the four preceding years (i.e., 2019, 2020, 2021, and 2022) as the estimation period. Each file concerns the results of a wind farm and a case study. The file name follows the pattern eval__.pickle, and the files are in Pickle format. Every file contains a Python dictionary with the following keys. The method labels used during the development process differ from those used in the paper. Please refer to the table below for the conversion of model labels. Label used in the paper Labels used in the scripts and files ENS-GBT-None direct_sppcm ENS-QGBT-None direct_probpcm HRES-QGBT-None AndradeBessa HRESc-QGBT-None AndradeBessa_cmp ENS-GBT-EMOS emos ENS-QGBT-bMM weighkd

This dataset contains the predicted time series and the forecast assessment conducted in the paper Seamless short- to mid-term probabilistic wind power forecasting.