search

AutoViML / Auto_ViML

Automatically Build Multiple ML Models with a Single Line of Code. Created by Ram Seshadri. Collaborators Welcome. Permission Granted upon Request.
Apache License 2.0
526 stars 102 forks source link

How are you handling preprocessing steps during prediction? #4

Closed deneshkumar closed 4 years ago

deneshkumar commented 5 years ago

Hi @rsesha

Currently Auto_ViML function is returning best model (XGB), features (Array), train metrics and test metrics. But how you suggesting to handling the preprocessing in the prediction dataset?

For Example, if you are applying LabelEncoding on a column inside the Auto_ViML function during training

rsesha commented 5 years ago

Denesh that will come in future versions. I have some code already ready. I just need to find time to integrate it and test. In the meantime, if you want to test the latest version, please do:

pip3 install --upgrade --force-reinstall --no-deps autoviml

This will take care of some bugs which crept in when I made the CatBoost upgrade. It has really messed up my code a bit.

Ram

On Wed, Dec 4, 2019 at 12:27 PM deneshkumar notifications@github.com wrote:

Hi @rsesha https://github.com/rsesha

Currently Auto_ViML function is returning best model (XGB), features (Array), train metrics and test metrics. But how you suggesting to handling the preprocessing in the prediction dataset?

For Example, if you are applying LabelEncoding on a column inside the Auto_ViML function during training

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/AutoViML/Auto_ViML/issues/4?email_source=notifications&email_token=AGEUZ7FOONX3LKKCXX67H4LQW7R7HA5CNFSM4JVMDZX2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4H6CKTZQ, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGEUZ7G3PPQEXU34LPMAXULQW7R7HANCNFSM4JVMDZXQ .

AutoViML commented 5 years ago

Actually this might work better for upgrading autoviml... pip3 install --upgrade --ignore-installed --no-deps autoviml Ram

On Wed, Dec 4, 2019 at 10:42 PM Ram Seshadri notifications@github.com wrote:

Denesh that will come in future versions. I have some code already ready. I just need to find time to integrate it and test. In the meantime, if you want to test the latest version, please do:

pip3 install --upgrade --force-reinstall --no-deps autoviml

This will take care of some bugs which crept in when I made the CatBoost upgrade. It has really messed up my code a bit.

Ram

On Wed, Dec 4, 2019 at 12:27 PM deneshkumar notifications@github.com wrote:

Hi @rsesha https://github.com/rsesha

Currently Auto_ViML function is returning best model (XGB), features (Array), train metrics and test metrics. But how you suggesting to handling the preprocessing in the prediction dataset?

For Example, if you are applying LabelEncoding on a column inside the Auto_ViML function during training

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub < https://github.com/AutoViML/Auto_ViML/issues/4?email_source=notifications&email_token=AGEUZ7FOONX3LKKCXX67H4LQW7R7HA5CNFSM4JVMDZX2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4H6CKTZQ , or unsubscribe < https://github.com/notifications/unsubscribe-auth/AGEUZ7G3PPQEXU34LPMAXULQW7R7HANCNFSM4JVMDZXQ

.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/AutoViML/Auto_ViML/issues/4?email_source=notifications&email_token=AMKBH6DE3TAT7EX56VUYHL3QXB2APA5CNFSM4JVMDZX2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEF7L46A#issuecomment-561954424, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMKBH6EIPCXVX3NLANH5BMLQXB2APANCNFSM4JVMDZXQ .

deneshkumar commented 5 years ago

Thanks for the quick response @rsesha

I understand this has great exciting ways to go. I would like to contribute to restructuring this code, feel free to pass on if you have specific coding task to be done. I will do it and raise pull request.

rsesha commented 5 years ago

Denesh: That would be awesome. I will ask you to start doing a CODE REVIEW with 2 goals in mind:

  1. Make minimal suggestions for change
  2. If you suggest, it can be to Clean up dead variables or speed up using map, apply etc instead of "For" loop Once that is done, I will ask you for more code changes. I expect this to be a long process but if you are willing to work with it over time, I am sure it can very rewarding. You can also send me emails via the email address prvided on my GitHub. Ram

On Wed, Dec 4, 2019 at 11:30 PM deneshkumar notifications@github.com wrote:

Thanks for the quick response @rsesha https://github.com/rsesha

I understand this has great exciting ways to go. I would like to contribute to restructuring this code, feel free to pass on if you have specific coding task to be done. I will do it and raise pull request.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/AutoViML/Auto_ViML/issues/4?email_source=notifications&email_token=AGEUZ7AJ4MMQH4BRN2TMWCDQXB7YFA5CNFSM4JVMDZX2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEF7OGYA#issuecomment-561963872, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGEUZ7G3YPSAC5YJNEHXPILQXB7YFANCNFSM4JVMDZXQ .