shenggong1996
commented
3 years ago Thanks Daniel! I will check the issue and re-do training if necessary.
Best, Sheng
From: @.> Sent: Thursday, September 30, 2021 4:03 PM To: @.> Cc: Sheng @.>; @.> Subject: [TRI-AMDD/htp_md] MAT-2976 (random forest model, fix test) (#26)
The issue was that int64 dtypes were filtered out prior to applying the model. In the Windows implementation of rdkit, more of the descriptors are int64 dtype. In the linux implementation still 2 descriptors are int64 and are also not included in the model.
Fix:
- include int64 dtypes
- manually exclude 2 descriptors which are currently not expected by the random forest model
@shenggong1996https://github.com/shenggong1996, Excluding these 2 descriptors from the training may or may not have been intended. In case that not, it might be worthwhile to try re-training the model including the two (see code for details on their names)
You can view, comment on, or merge this pull request online at:
https://github.com/TRI-AMDD/htp_md/pull/26
Commit Summary
- include int64 dtypeshttps://github.com/TRI-AMDD/htp_md/pull/26/commits/e8fbb75ce5903580c02e6adb48e8b32a66085850
- remove descriptors which the random forest model is not expectinghttps://github.com/TRI-AMDD/htp_md/pull/26/commits/fd8bbfddb1b31cc4901daa4db912ce2527ccca36
- temporary github workflows for triggering tests on push to branchhttps://github.com/TRI-AMDD/htp_md/pull/26/commits/0a60918d83c4ebfa2b36f4da9641ad5a04a8aae8
- fix syntax typohttps://github.com/TRI-AMDD/htp_md/pull/26/commits/deb8da6dafe302735c006c38cc8ad57e944b7ec0
- remove temporary test workflowshttps://github.com/TRI-AMDD/htp_md/pull/26/commits/6681c0a7edbcff3b9e0d95f69a36bb5ccaede205
File Changes
- M htpmd/ml_models/random_forest.pyhttps://github.com/TRI-AMDD/htp_md/pull/26/files#diff-a8c6bb159280b7bacc37d8d6bf7d48c45e59ed2dce2a53142df9b6d2cf10388a (3)
Patch Links:
- You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/TRI-AMDD/htp_md/pull/26, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AOXFYQ7IODZBLUNNYPYOV53UES7BVANCNFSM5FDDVWGA. Triage notifications on the go with GitHub Mobile for iOShttps://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Androidhttps://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.
danielschweigert-TRI
The issue was that int64 dtypes were filtered out prior to applying the model. In the Windows implementation of rdkit, more of the descriptors are int64 dtype. In the linux implementation still 2 descriptors are int64 and are also not included in the model.
Fix:
@shenggong1996, Excluding these 2 descriptors from the training may or may not have been intended. In case that not, it might be worthwhile to try re-training the model including the two (see code for details on their names)