Struggling with R.
My linear lm() Model contains Variables that are differenced via diff() and variables which are not differenced. The differenced variables are one observation shorter due to differencing. Therefore, the lm() gives the error message of different variable lengths.
My idea of a solution to this error is, to somehow define the variables as time series (which they are anyway, but R doesnt know that) and then tell the lm-Model exactly, which years to use (yearly data).
In my understanding, after differencing, a time-series looses its first observation, so, as I use the ts()-Funktion, I shall set a starting year one year later for the differenced function.
More concrete: lets say I imported the variables x and y then I go
dx<-diff(x, lag=1, differences=1)
while y remains the same
lm(y~dx)
will then produce the said error.
Lets say x and y both start in 1900. Then dx starts in 1901, so that lm has to start in 1901 for all Variables. My idea is, as stated above, to explicitly make both variables a time-series
tsdx<-ts(dx, frequency=1, start=1901)
tsy<-ts(y, frequency=1, start:1900)
and then somehow tell lm() to start with the year 1901.
Is this a good way to deal with these Problems? And how would I code the last step? Thanks alot!
