Regression using RBF kernel

Questions
3

Thanks for the suggestion. I implemented that, but now I don’t understand how to make the predictions because the number of predictions are always equal to the number of samples used for training.

I used the same rbf function and this is the PyMC3 implementation.

with pm.Model() as model:
    cov = tf.cholesky(rbf(X.get_value().astype(np.float32)))
    sess = tf.Session()
    with sess.as_default():
        cov = cov.eval()
    z = pm.MvNormal("z", mu=0, cov=cov.astype(np.float64), shape=100)
    sigma = pm.HalfCauchy("sigma", beta=10)
 
    likelihood = pm.Normal("y", mu=z, sd=sigma, observed=y)

Can you please help me with figuring out how to make the predictions?