cavalab / srbench

A living benchmark framework for symbolic regression
https://cavalab.org/srbench/
GNU General Public License v3.0
203 stars 74 forks source link

run the ML method ------metadata.yaml not found #168

Closed wencyxu closed 5 months ago

wencyxu commented 5 months ago
微信图片_20240124212312

When I run "python analyze.py /root/SP500/B1.tsv.gz -ml 'AdaBoostRegressor", it shows:Metadata file /root/SP500/metadata.yaml not found, skipping dataset /root/SP500/B1.tsv.gz. How do I run the ML method, please? Thanks a lot!!!!

lacava commented 5 months ago
  1. what version are you using? (recommend release)
  2. analyze.py expects a dataset directory, not a single data file. and it expects to submit to a cluster unless you pass --local flag.
wencyxu commented 5 months ago
微信图片_20240125092712

微信图片_20240125092719Thank you so much for receiving your reply ~ I have one more small question. As shown in the picture, I entered a dataset directory (containing tsv.gz and metadata.yaml), but it shows that the folder does not have a dataset.

folivetti commented 5 months ago

have you cloned the pmlb repository using git-lfs?

https://github.com/EpistasisLab/pmlb?tab=readme-ov-file#datasets

wencyxu commented 5 months ago

No, I haven't. Because I wanted to use my own dataset instead of PMLB.

lacava commented 5 months ago

https://github.com/cavalab/srbench/blob/f9fb2ba39d33ecabdc544f839008f13dca09ce29/experiment/analyze.py#L20 https://github.com/cavalab/srbench/blob/f9fb2ba39d33ecabdc544f839008f13dca09ce29/experiment/analyze.py#L21

pmlb/datasets looks like this: https://github.com/EpistasisLab/pmlb/tree/master/datasets

i.e., put your data in a data folder, and pass the parent folder.

if you aren't using the release version you may run into other issues.

wencyxu commented 5 months ago

Thanks a lot!!!😀I can run ML methods now.