fqr_prediction: Prediction intervals for functional quantile regression

In marcopalma3/neurofundata: Functional data analysis tools for neuroimaging data

Description

Return point predictions and prediction intervals for functional scalar-on-image quantile regression and outputs from FPCA.

Usage

fqr_prediction(
  data_projected_name,
  train_sub,
  test_sub,
  data_demo,
  pve_threshold,
  pred_table,
  qr_pen = "LASSO",
  qr_postLASSO = FALSE,
  lambda = NULL,
  tau_levels,
  return_func_coef = FALSE
)

Arguments

data_projected_name	text file with the smoothing projections for each statistical unit
train_sub	rownames of data_projected_name used to fit the FPCA
test_sub	rownames of data_projected_name for which to get the FPCA scores
data_demo	demographic data with the scalar outcome of interest
pred_table	table with predictions for each statistical unit
qr_pen	penalty type for quantile regression
qr_postLASSO	refit quantile regression with LASSO selected variables?
lambda	lambda parameter for penalised quantile regression
tau_levels	quantiles for which quantile regression is fitted
return_func_coef	return the functional coefficient?

Value

A list with the following elements

pred_table

table with predictions for each statistical unit

no_fpc

number of functional principal components selected

evec

Eigenimages

model_med_coef

Regression coefficient from median regression

model_lower_coef

Regression coefficient from regression with lower quantile

model_upper_coef

Regression coefficient from regression with lower quantile

lambda_min

Value of lambda for which model_med_coef is extracted

Author(s)

Marco Palma, [e-mail]

marcopalma3/neurofundata documentation built on Dec. 12, 2019, 5:29 a.m.