H2OTypeError: Argument `x` should be a None | integer | string | ModelBase | list(string | integer) | set(integer | string), got H2OFrame

Question

I working with the Titanic dataset and made some basic preprocessing (such as normalization, ohe, etc.).

Then, I tried to use H2O algorithm and got following error:

from h2o.estimators.gbm import H2OGradientBoostingEstimator
classifier  =  H2OGradientBoostingEstimator(nfolds =    5,
                                            ntrees =   15,
                                            seed   =    42,
                                            max_depth = 4)

classifier.train(predictors, target, training_frame = train_data)

H2OTypeError: Argument x should be a None | integer | string | ModelBase | list(string | integer) | set(integer | string), got H2OFrame

My target is train_data["Survived"].asfactor()

I tried to read to dataframe from file, instead of coverting the preprocessed df into H2OFrame but to no vail.

Any ideas would be appreciated.

Answer 1

It seems to me that you are passing frame instead of list of column names. Both x and y are supposed to be "pointers" to columns in the training_frame.

If you want to use all columns as predictors (all except the target), you can specify just the y parameter.

Something like the following should do the trick:

train_data["Survived"] = train_data["Survived"].asfactor()
classifier  =  H2OGradientBoostingEstimator(nfolds =    5,
                                            ntrees =   15,
                                            seed   =    42,
                                            max_depth = 4)

classifier.train(x=["pclass", "sex", "age",  ...], y="Survived", training_frame = train_data)