Dear Dr. Rasul,
I am writing to inquire about some unexpected results I encountered while attempting to replicate your research paper [MULTIVARIATE PROBABILISTIC TIME SERIES FORECASTING VIA CONDITIONED NORMALIZING FLOWS]. I have been working on a related research topic and found your paper to be very informative and helpful.
As part of my research, I attempted to replicate the results presented in your paper, specifically the calculation of the CRPS_sum metric. However, I noticed that after multiple attempts, the results(Solar and Exchange dataset) I obtained varied significantly and sometimes exceeded the variance reported in your paper.
I would appreciate it if you could help me understand why I might be obtaining such a large variance in the CRPS_sum metric when replicating your experiments. Are there any specific experimental conditions or parameters that could be contributing to this variability?
Additionally, I would like to know if there are any known limitations to the methods used in the paper or any other factors that could potentially impact the reproducibility of the results.
Thank you for taking the time to read this post. I look forward to your response.
Dear Dr. Rasul, I am writing to inquire about some unexpected results I encountered while attempting to replicate your research paper [MULTIVARIATE PROBABILISTIC TIME SERIES FORECASTING VIA CONDITIONED NORMALIZING FLOWS]. I have been working on a related research topic and found your paper to be very informative and helpful. As part of my research, I attempted to replicate the results presented in your paper, specifically the calculation of the CRPS_sum metric. However, I noticed that after multiple attempts, the results(Solar and Exchange dataset) I obtained varied significantly and sometimes exceeded the variance reported in your paper. I would appreciate it if you could help me understand why I might be obtaining such a large variance in the CRPS_sum metric when replicating your experiments. Are there any specific experimental conditions or parameters that could be contributing to this variability? Additionally, I would like to know if there are any known limitations to the methods used in the paper or any other factors that could potentially impact the reproducibility of the results.
Thank you for taking the time to read this post. I look forward to your response.
Alan