Question

我知道这个问题已存在于此：Creating an fts object in R with ftsa to functional data analysis of a time series

但缺少可重复的示例和答案，所以希望我的示例和数据得到答案。我的数据如下：

> data
         date  GDBR2  GDBR5 GDBR10 GDBR30
1  2015-10-15 -0.261 -0.032  0.550  1.310
2  2015-11-15 -0.370 -0.107  0.558  1.414
3  2015-12-15 -0.339 -0.049  0.641  1.411
4  2016-01-15 -0.394 -0.158  0.540  1.314
5  2016-02-15 -0.521 -0.312  0.237  0.928
6  2016-03-15 -0.453 -0.243  0.316  1.062
7  2016-04-15 -0.510 -0.382  0.127  0.805
8  2016-05-15 -0.513 -0.381  0.124  0.829
9  2016-06-15 -0.594 -0.483 -0.010  0.542
10 2016-07-15 -0.652 -0.551  0.006  0.530
11 2016-08-15 -0.613 -0.521 -0.074  0.433
12 2016-09-15 -0.640 -0.478  0.047  0.654

我能够创建声称为fts / zoo的fts1对象，但无法运行ftsm，因为is.fts返回FALSE

library(ftsa)
library(fts)

fts1 <- as.fts(data.frame(asofdate=data$date,data[,-1]))
class(fts1)
[1] "fts" "zoo"

ftsm(fts1, order=3)
Error in 1:ncol(y$y) : argument of length 0

is.fts(fts1)
[1] FALSE

对于如何创建ftsm函数中使用的fts对象的任何帮助表示赞赏。

这里输入数据：

> dput(data)
structure(list(date = structure(c(16723L, 16754L, 16784L, 16815L, 
16846L, 16875L, 16906L, 16936L, 16967L, 16997L, 17028L, 17059L
), class = "Date"), GDBR2 = c(-0.261, -0.37, -0.339, -0.394, 
-0.521, -0.453, -0.51, -0.513, -0.594, -0.652, -0.613, -0.64), 
    GDBR5 = c(-0.032, -0.107, -0.049, -0.158, -0.312, -0.243, 
    -0.382, -0.381, -0.483, -0.551, -0.521, -0.478), GDBR10 = c(0.55, 
    0.558, 0.641, 0.54, 0.237, 0.316, 0.127, 0.124, -0.01, 0.006, 
    -0.074, 0.047), GDBR30 = c(1.31, 1.414, 1.411, 1.314, 0.928, 
    1.062, 0.805, 0.829, 0.542, 0.53, 0.433, 0.654)), row.names = c(NA, 
-12L), class = "data.frame", .Names = c("date", "GDBR2", "GDBR5", 
"GDBR10", "GDBR30"))

Answer 1

有不同的包定义了一个类“fts”而你使用的是错误的。

library(ftsa)

#see the link in the documentation for the correct constructor
fts1 <- rainbow::fts(x = data$date, y = as.matrix(data[,-1]))

#apparently colnames must be numbers of the given frequency
#this appears to be at least an documentation bug
#but I know nothing about this kind of analysis
colnames(fts1$y) <- seq_along(colnames(fts1$y))

ftsm(fts1, order=3)
#Functional time series model
#
#Call: ftsm(y = fts1, order = 3) 
#
#Main effects: Mean 
#Order of interaction: 3
#   y: as.matrix(data[, -1]) 
#   x: data$date

在R中创建fts对象

1 个答案: