def predict(theta, X):
"""
Return predictions for a set of test scores.
"""
p = sigmoid(np.dot(X_array, theta)) >= 0.5
return p
# Newton-CG works well even without the Hessian.
# Per %timeit, about 3x faster than Nelder-Mead.
result_Newton_CG = minimize(lambda t: costFunction(t, X_array, y_array),
initial_theta,
method='Newton-CG',
jac=True)
# Print theta to screen
print 'Cost at theta found by Nelder-Mead: %f' % result_Nelder_Mead['fun']
print 'theta:', result_Nelder_Mead['x']
print 'Cost at theta found by Newton-CG: %f' % result_Newton_CG['fun']
print 'theta:', result_Newton_CG['x']
print ''
theta = result_Nelder_Mead['x']
# Plot Boundary
plotData(X, y, theta)
## ============== Part 4: Predict and Accuracies ==============
# Predict probability for a student with score 45 on exam 1
# and score 85 on exam 2
prob = sigmoid(np.dot(np.array([1, 45, 85]), theta))
print 'For a student with scores 45 and 85, we predict an admission probability of',
print '%.1f%%' % (prob * 100)
# Compute accuracy on our training set
p = predict(theta, X_array)
print 'Train Accuracy: %.1f%%' % ((p == y_array).mean() * 100)
