Question in calculating class labels for Table 3 in Paper
kmualim opened this issue · comments
Hi,
Thank you for releasing the code! I'm trying to reproduce the results in the paper and am wondering about the raw counts you've obtained in Table 3 of your paper.
Using the data provided here: https://extremeweatherdataset.github.io/?fbclid=IwAR12R2V5JZ8EbInNsnHazIcP5OMnhr_iiPkNjFzOHN_MRdjYCreUtxejDEo, I downloaded the 1979 and 1981 datasets and got the following class distribution.
1979 Dataset:
Tropical Cyclone: 0.442.738478, Total Count: 3190
Extra-Tropical Cyclones: 0.446.047696, Total Count: 3437
Tropical Depressions: 0.45.801179, Total Count: 433
United States Atmospheric Rivers: 0.45.412647, Total Count: 404
1984 Dataset:
Tropical Cyclone: 0.438.612004, Total Count: 2882
Extra-Tropical Cyclones: 0.444.386388, Total Count: 3313
Tropical Depressions: 0.49.338156, Total Count: 697
United States Atmospheric Rivers: 0.44.983923, Total Count: 372
All the values line up except for ETC where I got 3437 (1979 dataset) and 3313 (1984 dataset) while in the paper, it's 3510 (1979 dataset) and 3430 (1984 dataset)
-- is there a particular way you're calculating the numbers for ETC that would cause this discrepancy?
Thanks!
Hmmm interesting question. I believe I computed them using a metadata file and not the HDF5 files. The code I used for that is here: https://github.com/eracah/hur-detect/blob/master/notebooks/load_data/event_counter.ipynb
Lemme see if I can find that metadata file
Also, sorry for the late response and thanks for your interest! I apologize for the messy code and my fuzzy memory of this project! Let me know if you are still working on this and if you need any help
Hi yes, thank you for the response! Would you happen to also have the model weights available? :-)