Using 6 columns of time series data, conduct pairwise analysis on all possible iterations (in R)
up vote
0
down vote
favorite
I have a dataframe mydf containing 6 columns of time series data. I want to calculate the correlation of all this data, which can easily be done through cor(mydf) . However, I want to multiply each of the correlations by the square root of the relevant long-run variances (I adopt an arbitrary autocorrelation lag of 5) of each pairwise column. To demonstrate, val = cor(mydf[,1], mydf[,2]) cov_temp1 = acf(mydf[,1], type = "covariance", lag.max = 5, plot = FALSE, na.action = na.pass)$acf cov_temp2 = acf(mydf[,2], type = "covariance", lag.max = 5, plot = FALSE, na.action = na.pass)$acf s.e. = sqrt((cov_temp1[1]+2*sum(cov_temp1[-1]))/nrow(mydf) * (cov_temp2[1]+2*sum(cov_temp2[-1]))/nrow(mydf)) Then, the pairwise statistic for column 1 and 2 is val*s.e. . Assuming I have 6 colum...