NEWS

BayesGP 0.1.4

The initial location in f now could also support input of mean, min and max.
Update the function sample_fixed_effect to accept an input of M, denoting the number of posterior samples to take.
Update the method plot, to accept an argument of select, which selects which variable to show in the plot.
Create functions sample_sd that is used to extract posterior samples of the sd parameter, as well as its subsequent wrapper function para_samps which extracts posterior samples of any parameter.
custom_template now takes input in lower case rather than in upper case.

Fixed a bug in the structure of posterior samples when the model_fit object is of class "nlmb", now the object has posterior samples stored as a list object$samps$samps, which prevents the summary method from unexpected crashes.

var_plot and var_density are now called sd_plot and sd_density.
In model_fit, knots can now take a single numeric value, which will then be assumed as k.
The default order of iwp is assumed to be 2.
Plot now produce multiple figures of both the smoothing results and the SD parameters, which could be proceeded by pressing "next".

Fixed a bug in the global_poly_helper_sgp function and subsequently in the predict method that caused the prediction to fail when the smoothing variable is not pre-initialized manually. This should not affect models and predictions with manually initialized smoothing variables.

The definitions of initial_location being left, right and middle in IWP and sGP are not defined based on the region rather than the provided smoothing variable.