offsetFreq: Determine Offset Frequencies to include in model
In driegert/transfer: Frequency Domain Transfer Function Estimation Tools

Calculates the coherence between multiple inputs and an output and determines significant offset frequencies

offsetFreq(dat, responseName, blockSize = dim(dat)[1], overlap = 0,
  deltat = 1, nw = 4, k = 7, nFFT = NULL, mscCutoff = 0.2,
  freqRange = NULL, maxFreqOffset = 0, useZeroOffset = TRUE,
  nOffsetFreq = -1, calcType = 1, forward = 1)

`dat`	a `data.frame` whose columns contain the time-domain response and predictors
`responseName`	a `character` string giving the name of which column is the response.
`blockSize`	The length of a single block to use (if blocking)
`overlap`	A `numeric` value in the range [0, 1) indicating the proporation of overlap between neighbouring blocks.
`deltat`	The sampling rate in seconds.
`nw`	time-bandwidth parameter for multitaper
`k`	number of tapers to use (k < 2*nw)
`nFFT`	the number of frequency bins to use (nFFT > 2*ndata)
`freqRange`	A vector with 2 elements containing the start and end frequencies (in Hz) over which to calculate the coherence.
`maxFreqOffset`	Every pair of frequencies between f1 (series 1) and f1 +/- maxFreqOffset (series 2) will be calculated (f1 + maxFreqOffset < nyquist)
`calcType`	An `integer` value indicating how averaging over blocks should be performed: 1 - calculate the MSC on each block, then average; 2 - calculate the cross and auto spectra on each block, average each quantity across blocks, then calculate the coherency; 3 - calculate the coherency on each block, then average 4 - minimum MSC across blocks
`forward`	An `integer` indicating whether the forward (1) or reverse (0) coherence should be calculated.