Frequency parameter of `ts()` and `findfrequency()` in R Is the result obtained in findfrequency() function of forecast package and the frequency parameter of ts() the same?
 A: It probably should be, but there are no guarantees that it has to be.
findfrequency will return the period with the maximum spectral amplitude of the signal provided as an input. It is relative straight-forward calculation using the results from spec.ar where we estimate the spectral density associated with each candidate frequency of a time-series model from auto-regressive fit. Assuming that our raw signal, x, having assumed frequency 1, we use it as input to spec.ar:
specARresults <- spec.ar(x)

then then results from findfrequency are simply:
1 / specARresults$freq[which.max(specARresults$spec)]

i.e. the inverse of the frequency having the highest spectral density. And as the inverse of a frequency is a period we get our result. :) (Side-note: Why findfrequency returns a period eludes me.) 
We can theoretically use this result as input to ts's argument about frequency but we do not have to. The frequency  we use to ts is probably better to relate to our actual problem. For example, we might have daily expenditure data that happen to have a strong weekly pattern (cause people spend more on weekends). If we do not care for a weekly pattern though, it might be more relevant for us to analyse them assuming a period of 365 days.  
