# Gaussian Process Regression — PyMC3 3.1rc3 documentation (2023)

Gaussian Process regression is a non-parametric approach to regressionor data fitting that assumes that observed data points $$y$$ aregenerated by some unknown latent function $$f(x)$$. The latentfunction $$f(x)$$ is modeled as being multivariate normallydistributed (a Gaussian Process), and is commonly denoted

\begin{equation}f(x) \sim \mathcal{GP}(m(x;\theta), \, k(x, x';\theta)) \,.\end{equation}

$$m(x ; \theta)$$ is the mean function, and$$k(x, x' ;\theta)$$ is the covariance function. In manyapplications, the mean function is set to $$0$$ because the data canstill be fit well using just covariances.

$$\theta$$ is the set of hyperparameters for either the mean orcovariance function. These are the unknown variables. They are usuallyfound by maximizing the marginal likelihood. This approach is muchfaster computationally than MCMC, but produces a point estimate,$$\theta_{\mathrm{MAP}}$$.

The data in the next two examples is generated by a GP with noise thatis also gaussian distributed. In sampling notation this is,

\begin{equation}\begin{aligned}y & = f(x) + \epsilon \\f(x) & \sim \mathcal{GP}(0, \, k(x, x'; \theta)) \\\epsilon & \sim \mathcal{N}(0, \sigma^2) \\\sigma^2 & \sim \mathrm{Prior} \\\theta & \sim \mathrm{Prior} \,.\end{aligned}\end{equation}

With Theano as a backend, PyMC3 is an excellent environment fordeveloping fully Bayesian Gaussian Process models, particularly when aGP is component in a larger model. The GP functionality of PyMC3 ismeant to be lightweight, highly composable, and have a clear syntax.This example is meant to give an introduction to how to specify a GP inPyMC3.

In :
%matplotlib inlineimport matplotlib.pyplot as pltimport matplotlib.cm as cmapcm = cmap.infernoimport numpy as npimport scipy as spimport theanoimport theano.tensor as ttimport theano.tensor.nlinalgimport syssys.path.insert(0, "../../..")import pymc3 as pm

## Example 1: Non-Linear Regression¶

This is an example of a non-linear fit in a situation where there isn’tmuch data. Using optimization to find hyperparameters in this situationwill greatly underestimate the amount of uncertainty if using the GP forprediction. In PyMC3 it is easy to be fully Bayesian and use MCMCmethods.

We generate 20 data points at random x values between 0 and 3. Thetrue values of the hyperparameters are hardcoded in this temporarymodel.

In :
np.random.seed(20090425)n = 20X = np.sort(3*np.random.rand(n))[:,None]with pm.Model() as model: # f(x) l_true = 0.3 s2_f_true = 1.0 cov = s2_f_true * pm.gp.cov.ExpQuad(1, l_true) # noise, epsilon s2_n_true = 0.1 K_noise = s2_n_true**2 * tt.eye(n) K = cov(X) + K_noise# evaluate the covariance with the given hyperparametersK = theano.function([], cov(X) + K_noise)()# generate fake data from GP with white noise (with variance sigma2)y = np.random.multivariate_normal(np.zeros(n), K)
In :
fig = plt.figure(figsize=(14,5)); ax = fig.add_subplot(111)ax.plot(X, y, 'ok', ms=10);ax.set_xlabel("x");ax.set_ylabel("f(x)"); Since there isn’t much data, there will likely be a lot of uncertaintyin the hyperparameter values.

• We assign prior distributions that are uniform in log space, suitablefor variance-type parameters. Each hyperparameter must at least beconstrained to be positive valued by its prior.
• None of the covariance function objects have a scaling coefficientbuilt in. This is because random variables, such as s2_f, can bemultiplied directly with a covariance function object,gp.cov.ExpQuad.
• The last line is the marginal likelihood. Since the observed data$$y$$ is also assumed to be multivariate normally distributed,the marginal likelihood is also multivariate normal. It is obtainedby integrating out $$f(x)$$ from the product of the datalikelihood $$p(y \mid f, X)$$ and the GP prior$$p(f \mid X)$$,

\begin{equation}p(y \mid X) = \int p(y \mid f, X) p(f \mid X) df\end{equation}

• The call in the last line f_cov.K(X) evaluates the covariancefunction across the inputs X. The result is a matrix. The sum ofthis matrix and the diagonal noise term are used as the covariancematrix for the marginal likelihood.
In :
Z = np.linspace(0,3,100)[:,None]with pm.Model() as model: # priors on the covariance function hyperparameters l = pm.Uniform('l', 0, 10) # uninformative prior on the function variance log_s2_f = pm.Uniform('log_s2_f', lower=-10, upper=5) s2_f = pm.Deterministic('s2_f', tt.exp(log_s2_f)) # uninformative prior on the noise variance log_s2_n = pm.Uniform('log_s2_n', lower=-10, upper=5) s2_n = pm.Deterministic('s2_n', tt.exp(log_s2_n)) # covariance functions for the function f and the noise f_cov = s2_f * pm.gp.cov.ExpQuad(1, l) y_obs = pm.gp.GP('y_obs', cov_func=f_cov, sigma=s2_n, observed={'X':X, 'Y':y})
In :
with model: trace = pm.sample(2000)
Auto-assigning NUTS sampler...Initializing NUTS using advi...Average ELBO = -22.818: 100%|██████████| 200000/200000 [00:43<00:00, 4605.11it/s]Finished [100%]: Average ELBO = -22.857100%|██████████| 2000/2000 [00:18<00:00, 109.41it/s]

The results show that the hyperparameters were recovered pretty well,but definitely with a high degree of uncertainty. Lets look at thepredicted fits and uncertainty next using samples from the fullposterior.

In :
pm.traceplot(trace[1000:], varnames=['l', 's2_f', 's2_n'], lines={"l": l_true, "s2_f": s2_f_true, "s2_n": s2_n_true}); The sample_gp function draws realizations of the GP from thepredictive distribution.

In :
with model: gp_samples = pm.gp.sample_gp(trace[1000:], y_obs, Z, samples=50, random_seed=42)
 0%| | 0/50 [00:00<?, ?it/s]/Users/fonnescj/anaconda3/envs/dev/lib/python3.6/site-packages/scipy/stats/_multivariate.py:533: RuntimeWarning: covariance is not positive-semidefinite. out = random_state.multivariate_normal(mean, cov, size)100%|██████████| 50/50 [00:03<00:00, 16.24it/s]
In :
fig, ax = plt.subplots(figsize=(14,5))[ax.plot(Z, x, color=cm(0.3), alpha=0.3) for x in gp_samples]# overlay the observed dataax.plot(X, y, 'ok', ms=10);ax.set_xlabel("x");ax.set_ylabel("f(x)");ax.set_title("Posterior predictive distribution"); ## Example 2: A periodic signal in non-white noise¶

This time let’s pretend we have some more complex data that we wouldlike to decompose. For the sake of example, we simulate some data pointsfrom a function that 1. has a fainter periodic component 2. has a lowerfrequency drift away from periodicity 3. has additive white noise

As before, we generate the data using a throwaway PyMC3 model. Weconsider the sum of the drift term and the white noise to be “noise”,while the periodic component is “signal”. In GP regression, thetreatment of signal and noise covariance functions is identical, so thedistinction between signal and noise is somewhat arbitrary.

In :
np.random.seed(200)n = 150X = np.sort(40*np.random.rand(n))[:,None]# define gp, true parameter valueswith pm.Model() as model: l_per_true = 2 cov_per = pm.gp.cov.Cosine(1, l_per_true) l_drift_true = 4 cov_drift = pm.gp.cov.Matern52(1, l_drift_true) s2_p_true = 0.3 s2_d_true = 1.5 s2_w_true = 0.3 periodic_cov = s2_p_true * cov_per drift_cov = s2_d_true * cov_drift signal_cov = periodic_cov + drift_cov noise_cov = s2_w_true**2 * tt.eye(n)K = theano.function([], signal_cov(X, X) + noise_cov)()y = np.random.multivariate_normal(np.zeros(n), K)

In the plot of the observed data, the periodic component is barelydistinguishable by eye. It is plausible that there isn’t a periodiccomponent, and the observed data is just the drift component and whitenoise.

In :
fig = plt.figure(figsize=(12,5)); ax = fig.add_subplot(111)ax.plot(X, y, '--', color=cm(0.4))ax.plot(X, y, 'o', color="k", ms=10);ax.set_xlabel("x");ax.set_ylabel("f(x)"); Lets see if we can infer the correct values of the hyperparameters.

In :
with pm.Model() as model: # prior for periodic lengthscale, or frequency l_per = pm.Uniform('l_per', lower=1e-5, upper=10) # prior for the drift lengthscale hyperparameter l_drift = pm.Uniform('l_drift', lower=1e-5, upper=10) # uninformative prior on the periodic amplitude log_s2_p = pm.Uniform('log_s2_p', lower=-10, upper=5) s2_p = pm.Deterministic('s2_p', tt.exp(log_s2_p)) # uninformative prior on the drift amplitude log_s2_d = pm.Uniform('log_s2_d', lower=-10, upper=5) s2_d = pm.Deterministic('s2_d', tt.exp(log_s2_d)) # uninformative prior on the white noise variance log_s2_w = pm.Uniform('log_s2_w', lower=-10, upper=5) s2_w = pm.Deterministic('s2_w', tt.exp(log_s2_w)) # the periodic "signal" covariance signal_cov = s2_p * pm.gp.cov.Cosine(1, l_per) # the "noise" covariance drift_cov = s2_d * pm.gp.cov.Matern52(1, l_drift) y_obs = pm.gp.GP('y_obs', cov_func=signal_cov + drift_cov, sigma=s2_w, observed={'X':X, 'Y':y})
In :
with model: trace = pm.sample(2000, step=pm.NUTS(integrator="two-stage"), init=None)
100%|██████████| 2000/2000 [39:31<00:00, 1.67s/it]
In :
pm.traceplot(trace[1000:], varnames=['l_per', 'l_drift', 's2_d', 's2_p', 's2_w'], lines={"l_per": l_per_true, "l_drift": l_drift_true, "s2_d": s2_d_true, "s2_p": s2_p_true, "s2_w": s2_w_true}); Some large samples make the histogram of s2_p hard to read. Below isa zoomed in histogram.

In :
ax.get_ybound()
Out:
(0.0, 525.0)
In :
fig = plt.figure(figsize=(12,6)); ax = fig.add_subplot(111)ax.hist(trace['s2_p', 1000:], 100, range=(0,4), color=cm(0.3), ec='none');ax.plot([0.3, 0.3], [0, ax.get_ybound()], "k", lw=2);ax.set_title("Histogram of s2_p");ax.set_ylabel("Number of samples");ax.set_xlabel("s2_p"); Comparing the histograms of the results to the true values, we can seethat the PyMC3’s MCMC methods did a good job estimating the true GPhyperparameters. Although the periodic component is faintly apparent inthe observed data, the GP model is able to extract it with highaccuracy.

In :
Z = np.linspace(0, 40, 100).reshape(-1, 1)with model: gp_samples = pm.gp.sample_gp(trace[1000:], y_obs, Z, samples=50, random_seed=42, progressbar=False)
In :
fig, ax = plt.subplots(figsize=(14,5))[ax.plot(Z, x, color=cm(0.3), alpha=0.3) for x in gp_samples]# overlay the observed dataax.plot(X, y, 'o', color="k", ms=10);ax.set_xlabel("x");ax.set_ylabel("f(x)");ax.set_title("Posterior predictive distribution"); Top Articles
Latest Posts
Article information

Author: Stevie Stamm

Last Updated: 13/06/2023

Views: 5588

Rating: 5 / 5 (80 voted)

Author information

Name: Stevie Stamm

Birthday: 1996-06-22

Address: Apt. 419 4200 Sipes Estate, East Delmerview, WY 05617

Phone: +342332224300