search

jclevesque / bagged-budget-svms

Ensembles of Budgeted Kernel Support Vector Machines
MIT License
7 stars 1 forks source link

error in budgetedsvm #2

Open manvirvirk opened 3 years ago

manvirvirk commented 3 years ago

i m getting following error (esvm) D:\bagged-budget-svms-master>python setup.py install running install running build running build_py Compiling budgetedsvm toolbox.

error: [WinError 2] The system cannot find the file specified how to install budgetedsvm? THANKS

jclevesque commented 3 years ago

It seems you are using windows. I rely on some Unix tools such as make. You'd probably be better off using a vanilla machine learning model in scikit-learn.

manvirvirk commented 3 years ago

yes i m using windows, so is there any option as i don't have linux VM?

jclevesque commented 3 years ago

You can get the budgeted svm toolbox embedded in here to compile and then fix the bindings to call it in Windows. But it also depends on why you need this. If you want to compare ensembles of budgeted SVMs it might make sense to try to fix this, but if you just want any ensemble classifier there are many options out there.

manvirvirk commented 3 years ago

I want ensemble of SVM classifiers to classify features.

On Thu, Sep 24, 2020 at 5:45 PM Julien-Charles Lévesque < notifications@github.com> wrote:

You can get the budgeted svm toolbox embedded in here to compile and then fix the bindings to call it in Windows. But it also depends on why you need this. If you want to compare ensembles of budgeted SVMs it might make sense to try to fix this, but if you just want any ensemble classifier there are many options out there.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/jclevesque/bagged-budget-svms/issues/2#issuecomment-698305661, or unsubscribe https://github.com/notifications/unsubscribe-auth/ANAEAQDMGKLIL2QUK6U24ZTSHMZ5TANCNFSM4RV5EIIA .

jclevesque commented 3 years ago

If your problem is not large scale you can do that with scikit learn, build a wrapper around the SVM classifier, train multiple SVMs on bagged subsamples (or subspaces, feature-wise) of the training set. I'd look into Gradient boosted trees, random forests and XGboost, those are much more mature now and can work for most use cases.

On Fri, 25 Sep 2020 at 04:29, manvirvirk notifications@github.com wrote:

I want ensemble of SVM classifiers to classify features.

On Thu, Sep 24, 2020 at 5:45 PM Julien-Charles Lévesque < notifications@github.com> wrote:

You can get the budgeted svm toolbox embedded in here to compile and then fix the bindings to call it in Windows. But it also depends on why you need this. If you want to compare ensembles of budgeted SVMs it might make sense to try to fix this, but if you just want any ensemble classifier there are many options out there.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub < https://github.com/jclevesque/bagged-budget-svms/issues/2#issuecomment-698305661 , or unsubscribe < https://github.com/notifications/unsubscribe-auth/ANAEAQDMGKLIL2QUK6U24ZTSHMZ5TANCNFSM4RV5EIIA

.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/jclevesque/bagged-budget-svms/issues/2#issuecomment-698798523, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAFEP2HBWU2T24SF3NYY2RTSHRIGZANCNFSM4RV5EIIA .