2. What happens if you omit the sub - command use = "pairwise . complete .obs " from cor ( ) and why ? The Autocorrelation Function (acf ) The lag k autocorrelation is usually denoted by P, = Corr(Y,, Y.) = Cov (Y, Y..)/Var(Y.). We have implicitly assumed that these are constant over time. This will follow provided the time series is stationary , specifically 2" order stationary, ie a. it has a well -defined , constant mean level ", b. a constant finite variance ),, and c. the covariances Y, = Cov (Y., Y.k) depend only on k and not on t. So it follows that the autocorrelations Px = Vr/Y, depend only on k and not on t. To save us the effort of calculating lots of these lagged correlations the acf( ) function calculates and plots them for us . act ( hare )