0

I am trying to use AutoSklearn with a specific list of algorithms

Logistic Regression
Random Forest
Gaussian NB
SVC
ADA
MLP

I know I can use these parameters.

mdl = autosklearn.classification.AutoSklearn2Classifier(
    include = {
         'classifier': ["random_forest", "gaussian_nb", "libsvm_svc", "adaboost"],
         'feature_preprocessor': ["no_preprocessing"]
    #K-Folds?
    },
    exclude=None,

I managed to find the code for these algorithms

Random Forest  ==> "random_forest"
Gaussian NB  ==> "gaussian_nb"
SVC  ==> "libsvm_svc"
ADA  ==> "adaboost"

but could not find the codes for

Logistic Regression
MLP

can anyone tell me what are these?

asmgx
  • 7,328
  • 15
  • 82
  • 143

1 Answers1

2

The documentation states that the strings used to identify estimators and preprocessors are the filenames without .py.

You can find here the model_id you are looking for here.

From the documentation MLP code is mlp, and Logistic Regression is not implemented. (see this issue for further information)

Therefore you should do as follows:

mdl = autosklearn.classification.AutoSklearn2Classifier(
    include = {
         'classifier': ["random_forest", "gaussian_nb", "libsvm_svc", "adaboost", 'mlp'],
         'feature_preprocessor': ["no_preprocessing"]
    },
    exclude=None
)
Antoine Dubuis
  • 4,974
  • 1
  • 15
  • 29
  • Thanks @Antoine, Do you know how can I specify the HyperParameters in AutoSklearn? I want to specify random_forst with Estimator = 1000 only or MLP with HiddenLayerSize = 100 only.. any idea how to do that? – asmgx Jan 19 '22 at 04:54