Created
August 29, 2017 22:58
-
-
Save iandewancker/9f86267e54ee6342ee58fd84fac1ba87 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
X_test = np.vstack([test_x1, test_x2, test_x3, test_x4, test_x5, test_x6]) | |
y_test = np.array([1.0, 1.0, 1.0, 1.0, -1.0, -1.0]) | |
clf.model.coef_ = clf.model.coef_.reshape(1,clf.model.coef_.shape[0]) | |
# hack to set the classes | |
try: | |
clf.model.fit([],[0,1]) | |
except: | |
pass | |
# gradient update step simulated | |
alpha = 0.00125 | |
w_u = clf.model.coef_.flatten() | |
b_u = clf.model.intercept_ | |
for idx in xrange(X_test.shape[0]): | |
w_u = w_u - Log.dloss(Log(),clf.model.decision_function(X_test[idx]),y_test[idx])*X_test[idx]*alpha | |
b_u = b_u - Log.dloss(Log(),clf.model.decision_function(X_test[idx]),y_test[idx])*alpha | |
clf2 = LogisticRegression() | |
clf2.coef_ = w_u.reshape(1,w_u.shape[0]) | |
clf2.intercept_ = b_u | |
try: | |
clf2.fit([],[0,1]) | |
except: | |
pass | |
print "NEW", clf2.predict_proba(X_test) | |
print "OLD", clf.model.predict_proba(X_test) | |
print "NEW", clf2.predict(X_test) | |
print "OLD", clf.model.predict(X_test) | |
print "NEW", clf2.coef_, clf2.intercept_ | |
print "OLD",clf.model.coef_, clf.model.intercept_ |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment