This repository contains R code to reproduce the results presented in the paper
Allen, S., Ginsbourger, D. and Ziegel, J. (2024). Efficient pooling of predictions via kernel embeddings. ArXiv preprint. arxiv.org/abs/2411.16246
The paper concerns methods to efficiently combine probabilistic forecasts via kernel mean embeddings.
The results can be reproduced and saved by sourcing the main.R file. The saved results are stored in the Results folder. Functions to perform the RKHS-based forecast combination methods can be found in the utility_funcs.R file.
We are extremely grateful to MeteoSwiss (The Swiss Federal Office of Meteorology and Climatology MeteoSwiss) for providing the wind speed data used in this study, and for allowing it to be shared publicly. The data is available in the Data folder. A separate RDS file is provided for the forecasts and observations at each lead time. Files containing the station info, such as longitude, latitude, and height, are also available.