Due to the mysterious TypeError, Scikit-learn GridSearchCV cannot fit EM models using silhouette_score
The following code causes: TypeError: __call__() takes at least 4 arguments (3 given).
I’ve instantiated a clustering classifier and a scoring method created that fits the cluster. I provide a simple dataset for fitting and a dictionary of parameters for grid search. It’s hard for me to see where there are errors, and backtracking doesn’t help.
from sklearn.mixture import GaussianMixture
from sklearn.model_selection import GridSearchCV
from sklearn.metrics import silhouette_score, make_scorer
parameters = {'n_components': range(1, 6), 'covariance_type': ['full', 'tied', 'diag', 'spherical']}
silhouette_scorer = make_scorer(silhouette_score)
gm = GaussianMixture()
clusterer = GridSearchCV(gm, parameters, scoring=silhouette_scorer)
clusterer.fit(data)
Backtracking is mysterious, as far as I can tell, I’m following syntax and workflow described in GridSearchCV’s sklearn documentation. Exactly. What am I doing wrong here that causes this error?
The data content is as follows:
Dimension 1 Dimension 2
0 -0.837489 -1.076500
1 1.746697 0.193893
2 -0.141929 -2.772168
3 -2.809583 -3.645926
4 -2.070939 -2.485348
.. ... ...
401 -0.477716 -0.347241
402 0.742407 0.005890
403 -2.152810 5.385891
404 -0.074108 -1.691082
405 0.555363 -0.002872
416 -1.597249 -0.804744
Here are the last few lines of the traceback:
/usr/local/lib/python2.7/site-packages/sklearn/externals/joblib/parallel.pyc in __call__(self)
129
130 def __call__(self):
--> 131 return [func(*args, **kwargs) for func, args, kwargs in self.items]
132
133 def __len__(self):
/usr/local/lib/python2.7/site-packages/sklearn/model_selection/_validation.pyc in _fit_and_score(estimator, X, y, scorer, train, test, verbose, parameters, fit_params, return_ train_score, return_parameters, return_n_test_samples, return_times, error_score)
258 else:
259 fit_time = time.time() - start_time
--> 260 test_score = _score(estimator, X_test, y_test, scorer)
261 score_time = time.time() - start_time - fit_time
262 if return_train_score:
/usr/local/lib/python2.7/site-packages/sklearn/model_selection/_validation.pyc in _score(estimator, X_test, y_test, scorer)
284 """Compute the score of an estimator on a given test set."""
285 if y_test is None:
--> 286 score = scorer(estimator, X_test)
287 else:
288 score = scorer(estimator, X_test, y_test)
TypeError: __call__() takes at least 4 arguments (3 given)
Solution
Well, the
problem is, you used the wrong function as an argument to make_scorer
. documentation for make_scorer
Say:
score_func – Score function (or loss function) with signature score_func(y_true, y_pred, **kwargs)
And you pass silhouette_score
to it, one of them signature (X, labels, metric='euclidean' ...)
Obviously does not meet the requirements of make_scorer
, so an error is reported.
Try changing it to a different metric to resolve the error.