Speeding up the watcher #40
Description
- would a smaller model be faster to run? a 5GB random forest seems unneccessarily large when it should be adequate to describe a particle with only three or four parameters / "dimensions"
- testing on this large file:
https://citprodflowcytosa.blob.core.windows.net/hdduploadnov2024/data_transfer/surveys/CEND05_23/image_test1%202023-04-24%2011h47.cyz
the cyz2json step was taking about 9 minutes. We don't need the images in the json file, so it would be ideal if cyz2json had a "noimages" flag because that must be what is taking most of the time in this case. - We could have a user option to ditch the .html plot file which may marginally speed things up