Moving Functions
apply summary functions on a moving/rolling basis
moving.mean(x, periods) moving.sum(x, periods) moving.max(x, periods) moving.min(x, periods) moving.sd(x, periods) moving.rank(x, periods) moving.cor(x, y, periods) moving.cov(x, y, periods) cor.by.row(x,y)
x |
An Fts object |
y |
An Fts object |
periods |
integer: number of periods in window |
apply a function that takes a vector and returns a scalar on a rolling basis to an fts object.
For cor.by.row, the indicator is not rolling, but is the result of the application of the cor function to matching rows of x and y.
asking for a window larger than the number of rows of the fts object will result in an fts of all NA w/ the same number of rows as the input
for functions that take two fts objects the date intersection is taken before the window function is applied
an fts object
Whit Armstrong
x <- fts(index=seq(from=Sys.Date(),by="days",length.out=100),data=1:100) y <- fts(index=seq(from=Sys.Date(),by="days",length.out=100),data=1:100) x.mean <- moving.mean(x,20) x.sum <- moving.sum(x,20) x.prod <- moving.product(x,20) x.max <- moving.max(x,20) x.min <- moving.min(x,20) x.sd <- moving.sd(x,20) x.rank <- moving.rank(x,20) ## take only odd rows ## to illustrate that teh correlation and covariance ## will only be calculated for the intersection of the dates y <- y[(1:nrow(y))%%2 == 1] xy.cor <- moving.cor(x, y, 20) xy.cov <- moving.cov(x, y, 20)
Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.