I am new to using PyMC3 and I am still learning, but very enthusiastic about the initial results. I have been working for many years with artificial neural networks and always with non-linear multivariate data. My initial tests with PyMC3 were done using the following model (the main line):
mu = pm.Deterministic('mu', a + B[0] * X1 + B[1] * X2 + B[2] * X3 + B[3] * X4 + B[4] * X5)
Considering an output vector Y and an input matrix X with 5 columns, each column Xi representing a different variable. This model above using Gaussian (Normal) distributions provides reasonable results, but I would like to know if there is a specific approach to non-linear mutivated data. I did not find similar questions to get my doubts. Thanks for any feedback!