MLP Parameter tuning - gridsearchCV cannot fit?

Question

I'm making an MLP classifier for binomial classification from 145 features.

I want to get the best parameters on my MLP classifier to get a better prediction so I followed the answer to this question, which is to use gridsearchCV from sklearn. However, when I get to

clf.fit(DEAP_x_train, DEAP_y_train)

I get the ff error:

TypeError: '<=' not supported between instances of 'str' and 'int'

I checked my train data and both X and y have already been changed to become float data type, so I don't understand why I keep getting this response.

For reference, I have 145 features to predict a binary response, and this is my parameter_space:

parameter_space = {
    'hidden_layer_sizes': [(368,), (555,), (100,)],
    'activation': ['identity', 'logistic', 'relu'],
    'solver': ['sgd', 'adam'],
    'alpha': [0.0001, 0.05],
    'learning_rate': ['constant','adaptive'],
    'max_iter': ['200', '1000', '5000', '10000']
}

I've tried to search for other cases online that this error response is shown when trying to fit, but I haven't come across any.

Would appreciate any response. Thanks!

Can you share the code where are the less than or equal to comparison appears? — grldsndrs, Jul 14 '19 at 13:13

score 2 · Accepted Answer · answered Jul 14 '19 at 13:24

2

Your max iteration values are strings.

max_iter': ['200', '1000', '5000', '10000']

Try

max_iter': [200, 1000, 5000, 10000] }

answered Jul 14 '19 at 13:24

grldsndrs

567
4
11

Thanks so much!!! Such an elementary mistake on my part :( – Lasmyr Jul 15 '19 at 01:00

MLP Parameter tuning - gridsearchCV cannot fit?

1 Answers1