TPOT error in python cannot set using a slice indexer with a different length

Question

I'm trying to run tpot to optimize hyperparameters of a random forest using genetic algorithms. I am receiving an error and am not quite sure how to fix it. Below is the essential code I'm using.

from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import RandomizedSearchCV
from sklearn.model_selection import train_test_split
from tpot import TPOTClassifier

X = my_df_features
y = my_df_target

X_train, X_test, y_train, y_test = train_test_split(X,y, random_state=42)

model_parameters = {'n_estimators': [100,200],
                      "max_depth" : [None, 5, 10],
                      "max_features" : [10]}  

# This seems to work perfectly fine when I run it
# model_tuned = GridSearchCV(RandomForestClassifier(),model_parameters, cv=5)

# This does not seem to work 
model_tuned = TPOTClassifier(generations= 2, population_size= 2, offspring_size= 2,
                                      verbosity= 2, early_stop= 10,
                                      config_dict=
                                      {'sklearn.ensemble.RandomForestClassifier': model_parameters}, 
                                      cv = 5)
 

model_tuned.fit(X_train,y_train)

When using TPOT (as opposed to RandomForest), the last line above produces the following error:

ValueError: cannot set using a slice indexer with a different length than the value"

Please do not hide error messages inside code (edited). You should post the full error trace separately. — desertnaut, Dec 01 '20 at 11:46

Nikaido · Accepted Answer · 2020-12-01T13:43:44.263

I tried tpot with the iris dataset and I did get no error

from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import RandomizedSearchCV
from sklearn.model_selection import train_test_split
from tpot import TPOTClassifier
from sklearn import datasets
iris = datasets.load_iris()
X = iris.data 
y = iris.target

X_train, X_test, y_train, y_test = train_test_split(X,y, random_state=42)

model_parameters = {'n_estimators': [100,200],
                      "max_depth" : [None, 5, 10],
                      "max_features" : [len(X_train[0])]}  


model_tuned = TPOTClassifier(generations= 2, 
                             population_size= 2, 
                             offspring_size= 2,
                             verbosity= 2, 
                             early_stop= 10, 
                             config_dict={'sklearn.ensemble.RandomForestClassifier': 
                             model_parameters}, 
                             cv = 5)
 

model_tuned.fit(X_train,y_train)

I think there is something wrong in the shape or type of your dataset

Maybe due to the fact that you are using pandas DataFrames

Try to do this:

X = X.to_numpy
y = y.to_numpy

Thanks! Your comment actually led me to the solution. Turned out that it was an issue with the shape of the training and testing datasets. Specifically, there was a different number of classes in the training and testing datasets. — thehand0, Dec 01 '20 at 21:52

TPOT error in python cannot set using a slice indexer with a different length

1 Answers1