在 SVM 上出现错误“TypeError: an integer is required”

Question

I'm pretty new to coding so I get a bit confused about what my code actually means and dictates, so sorry if this is a dumb question or if I struggle to understand answers!

So here's the code I'm having issues with:

krn= ['linear', 'poly', 'rbf', 'sigmoid']
rng_C = np.arange(1,101,20)
rng_degree=np.arange(2,5)
rng_ga= ['auto', 'scale']
rng_m=np.arange(0.001,10,0.5)

best_score=0
for i in krn:
    for j in rng_C:
        for k in rng_ga:
            for m in rng_m:
                SVModel=SVC(kernel=i, C=j, degree=k, gamma=m)
                SVModel.fit(x_train, y_train)
                acc_score= accuracy_score(y_test, SVModel.predict(x_test))
                if best_score<acc_score:
                    best_score=acc_score
                    bi=i
                    bj=j
                    bk=k
                    bm=m
print(best_score, bi, bj, bk, bm)

The error screen I'm getting is:

TypeError                                 Traceback (most recent call last)
<ipython-input-9-5eb9a5279e65> in <module>
      5             for m in rng_m:
      6                 SVModel=SVC(kernel=i, C=j, degree=k, gamma=m)
----> 7                 SVModel.fit(x_train, y_train)
      8                 acc_score= accuracy_score(y_test, SVModel.predict(x_test))
      9                 if best_score<acc_score:

~\anaconda3\lib\site-packages\sklearn\svm\_base.py in fit(self, X, y, sample_weight)
    224 
    225         seed = rnd.randint(np.iinfo('i').max)
--> 226         fit(X, y, sample_weight, solver_type, kernel, random_seed=seed)
    227         # see comment on the other call to np.iinfo in this file
    228 

~\anaconda3\lib\site-packages\sklearn\svm\_base.py in _dense_fit(self, X, y, sample_weight, solver_type, kernel, random_seed)
    275         self.support_, self.support_vectors_, self._n_support, \
    276             self.dual_coef_, self.intercept_, self._probA, \
--> 277             self._probB, self.fit_status_ = libsvm.fit(
    278                 X, y,
    279                 svm_type=solver_type, sample_weight=sample_weight,

sklearn\svm\_libsvm.pyx in sklearn.svm._libsvm.fit()

TypeError: an integer is required

From what I've read so far, the issue seems to be that I'm somehow getting a string and it's getting input where I actually need an integer. I'm unsure of how to do that, and where I would add the potential code needed to fix the issue. If anyone has any insight I would greatly appreciate it!

Answer 1

You've mixed up rng_ga and rng_degree in the line for k in rng_ga: , so the degree value k becomes a string "auto" when passed to line SVModel=SVC(kernel=i, C=j, degree=k, gamma=m)