Model Training Performance

[elquintodev]

I tried to create a Random Forest model (tree=100, min split =2, max dept=5) with 5 features and trained it for 150,000 sample set. However, it's been more than 8 hours and haven't seen at least 1 tree being processed (log stuck at "Classifier Random Forest Building"). Tried to decrease the number of sample set to 10,000, but no luck after 6 hours. Decreased again down to 1,000 and that's the only time it processed and model was trained successfully (after around 1hr).

Is there anyway we can optimize and improve the performance when building/training the model for larger sample size?

My machine is running on 32gb ram, i7 12core 16 logical processors.

[MegaJoctan]

My apologies for the issue you are facing, I just realized the mistake I made programming the decision tree model, where I used recursive programming instead of vectorized operations which are suitable for machine learning. I will work to improve it shortly. Stay tuned, and leave this issue open. On Mon, May 12 2025 at 9:01 AM El Quinto ***@***.***> ***@***.***> wrote: I tried to create a Random Forest model (tree=100, min split =2, max dept=5) with 5 features and trained it for 150,000 sample set. However, it's been more than 8 hours and haven't seen at least 1 tree being processed (log stuck at "Classifier Random Forest Building"). Tried to decrease the number of sample set to 10,000, but no luck after 6 hours. Decreased again down to 1,000 and that's the only time it processed and model was trained successfully. Is there anyway we can optimize building/training the model for larger sample size? My machine is running on 32gb ram, i7 12core 16 logical processors. — Reply to this email directly, view it on GitHub <#10>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/APSQQFNVMNQLRRABWH45C5L26A2LFAVCNFSM6AAAAAB446UX3CVHI2DSMVQWIX3LMV43ERDJONRXK43TNFXW4OZYGMYTAOJQGI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[elquintodev]

Great! Thanks!