Last active
October 2, 2019 14:05
-
-
Save patricoferris/93dd9d1701458a0f656ac1d8f2da4dda 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
# First we calculate mu_s | |
alpha = 2. | |
K_xy = rbf_kernel(xs_train, xs_test, alpha) | |
K_yy = rbf_kernel(xs_test, xs_test, alpha) | |
K_xx = rbf_kernel(xs_train, xs_train, alpha) | |
np.random.seed(20) | |
mu_s = np.matmul(K_xy.T, np.matmul(np.linalg.pinv(K_xx), (ys_train))) | |
# Second we calculate sigma_s | |
sigma_s = K_yy - np.matmul(K_xy.T, np.matmul(np.linalg.pinv(K_xx), K_xy)) | |
# Now we draw from the prior by using the method for creating normally distributed points | |
L = np.linalg.cholesky(sigma_s + 0.00001*np.eye(len(xs_test))) | |
f_post = mu_s + np.dot(L, np.random.normal(size=(len(xs_test), 1))) | |
f_post = np.linalg.norm(og_temps) * f_post | |
fig, ax = plt.subplots(figsize=(10, 6)) | |
ax.plot(xs, og_temps, 'b.', markersize=10, label=u'True Points') | |
ax.plot(training_points, og_temps[training_points], 'r.', markersize=10, label=u'Observations') | |
ax.plot(xs_test, f_post) | |
ax.fill(np.concatenate([xs_test, xs_test[::-1]]), | |
np.concatenate([f_post - np.linalg.norm(og_temps) * 1.9600 * sigma_s, | |
(f_post + np.linalg.norm(og_temps) * 1.9600 * sigma_s)[::-1]]), | |
alpha=.5, fc='0.8', ec='None', label='95% confidence interval') | |
ax.set_title('Predicting New York Temperatures') | |
plt.show() | |
fig.savefig("nyc_temp_predictions") |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment