make_random_truth_generator

Generates reproducible random truth functions by sampling from P-spline priors.
Uses the same penalty structure as mgcv, so wiggliness maps to effective
degrees of freedom. Truth is in the B-spline span, avoiding approximation bias.

internal

Methods for regression for functional
data, including function-on-scalar, scalar-on-function, and
function-on-function regression. Some of the functions are applicable to
image data.

Julia Wrobel

refund

Regression with Functional Data

Jeff Goldsmith

Fabian Scheipl

Lei Huang

Chongzhi Di

Jonathan Gellar

Jaroslaw Harezlak

Bruce Swihart

Luo Xiao

Ciprian Crainiceanu

Yakuan Chen

Sonja Greven

Lan Huo

So Young Park

Ana-Maria Staicu

Erjia Cui

Ruonan Li

Zheyuan Li

make_random_truth_generator function

<dl><dt>yind</dt>
<dd>Response grid points.</dd>
<dt>xind</dt>
<dd>Functional covariate grid points.</dd>
<dt>z_ref</dt>
<dd>Reference z values for smooth terms.</dd>
<dt>wiggliness</dt>
<dd>Controls smoothness (default: 1). Higher = more wiggly.
Typical range: 0.001 (very smooth) to 10 (very wiggly).</dd>
<dt>k_truth</dt>
<dd>List with basis dimensions per term type. Defaults:
ff_s = 8, ff_t = 8, smooth_z = 8, smooth_t = 8,
linear = 8, intercept = 8, concurrent = 8.</dd></dl>

Arguments

Create random truth generator with P-spline prior — make_random_truth_generator

<dl>

<dt>yind</dt>
<dd>Response grid points.</dd>


<dt>xind</dt>
<dd>Functional covariate grid points.</dd>


<dt>z_ref</dt>
<dd>Reference z values for smooth terms.</dd>


<dt>wiggliness</dt>
<dd>Controls smoothness (default: 1). Higher = more wiggly.
Typical range: 0.001 (very smooth) to 10 (very wiggly).</dd>


<dt>k_truth</dt>
<dd>List with basis dimensions per term type. Defaults:
ff_s = 8, ff_t = 8, smooth_z = 8, smooth_t = 8,
linear = 8, intercept = 8, concurrent = 8.</dd>

</dl>

Create random truth generator with P-spline prior

make_random_truth_generator: Create random truth generator with P-spline prior

Description

Usage

Value

Arguments