R/fmri_ts_forecast.R

In TCIU: Spacekime Analytics, Time Complexity and Inferential Uncertainty

#' @title forecast the fMRI data based on the time series
#' @description a function to forecast the fMRI data based on the time series
#'
#' @param fmridata a 4D array contains information for the fMRI spacetime image. The data should only contain the magnitude for the fMRI image.
#' @param voxel_location a 3d array indicating the voxel location of the brain
#' @param cut breaking point of the time-series data. The default is 10. 
#'
#' @details The function \code{fmri_ts_forecast} is used to forecast with time series. It will fit the best ARIMA model to univariate time series from the input fMRI data.
#'
#' @author SOCR team <\url{http://socr.umich.edu/people/}>
#'
#' @return a figure forecasting the fMRI voxel with time series
#' @export
#' @import zoo plotly
#' @importFrom stats time
#' @importFrom forecast auto.arima forecast
#' 
#' @examples
#' fmri_generate = fmri_simulate_func(dim_data = c(64, 64, 40), mask = mask)
#' \donttest{
#' smoothmod <- GaussSmoothArray(fmri_generate$fmri_data, sigma = diag(3,3))
#' fmri_ts_forecast(smoothmod,c(41,44,33))
#' }

fmri_ts_forecast = function(fmridata,
                            voxel_location,
                            cut = 10) {
  fdata = fmridata[voxel_location[1], voxel_location[2], voxel_location[3], ]
  trainsize=floor(length(fdata)*0.8)
  data = detrend(fmridata[voxel_location[1], voxel_location[2], voxel_location[3], ], bp = seq(21, length(fdata), by = cut))
  trainData <- data[1:trainsize]
  testData <- ts(data[(trainsize+1):length(fdata)], start = (trainsize+1), end = length(fdata))
  arimaMod <- auto.arima(trainData, stepwise = FALSE, approximation = FALSE)
  TSplot_gen(trainsize, arimaMod,periods=length(fdata)-trainsize , ts_list = list(testData))
}