通过多个变化的变量和不平衡的时间将宽重塑为长

Question

我有两组不同的变量：年度百分比和年份。 Annualpercentage 从 1999 年开始到 2012 年结束，但年份从 1999 年开始到 2013 年。

countrylabel annualpercentageshare.1999 year1990 year1991 year1992
1      Austria                         NA       NA       NA       NA
2      Belgium                         NA       NA       NA       NA
3     Bulgaria                   48.20000       NA       NA       NA
4      Estonia                         NA       NA       NA       NA
5       France                   47.52853       NA       NA       NA
6      Germany                         NA       NA       NA       NA

类似这样的事情。

我已经尝试过这段代码：

merge_data2 <- reshape(merge_data2, varying = list(2:ncol(merge_data2)), 
                       v.names = c("percentageshare", "Year"),
                       idvar = "countrylabel", direction = "long", times = 1990:2013)

但我收到此错误消息：

“reshapeLong 中的错误（数据，idvar = idvar，timevar = timevar，变化 = 变化，： “长度（变化）”必须全部匹配“长度（次）””

编辑：我想要一个像这样的数据框：

countrylabel    time      annualpercentageshare        year
Austria          1990            NA                      NA
Austria          1991            NA                      NA

Answer 1

library(tidyr); library(dplyr)
df %>%
  gather(variable, value, -countrylabel) %>%
  separate("variable", into = c("stat", "time"), sep = -4) %>%
  spread(stat, value)

输出

   countrylabel time annualpercentageshare. year
1       Austria 1990                     NA   NA
2       Austria 1991                     NA   NA
3       Austria 1992                     NA   NA
4       Austria 1999                     NA   NA
5       Belgium 1990                     NA   NA
6       Belgium 1991                     NA   NA
7       Belgium 1992                     NA   NA
8       Belgium 1999                     NA   NA
9      Bulgaria 1990                     NA   NA
10     Bulgaria 1991                     NA   NA
11     Bulgaria 1992                     NA   NA
12     Bulgaria 1999               48.20000   NA
13      Estonia 1990                     NA   NA
14      Estonia 1991                     NA   NA
15      Estonia 1992                     NA   NA
16      Estonia 1999                     NA   NA
17       France 1990                     NA   NA
18       France 1991                     NA   NA
19       France 1992                     NA   NA
20       France 1999               47.52853   NA
21      Germany 1990                     NA   NA
22      Germany 1991                     NA   NA
23      Germany 1992                     NA   NA
24      Germany 1999                     NA   NA

Answer 2

reshape

喜欢

"."

，所以首先我们将一个插入到

year*

变量中。

names(d) <- gsub("year", "year.", names(d))

现在我们给出了

reshape

缺失的列和

order

,

d$annualpercentage.2002 <- NA
d$year.1999 <- NA
d <- d[c(1, order(names(d)[-1]) + 1)]

您的想法通过在列表中的

varying

中定义不同的列排序来实现：

res <- reshape(d, varying=list(2:5, 6:9), direction="long", idvar="countrylabel", 
               times=1999:2002, v.names=c("annualpercentage", "year"))
res
#                  countrylabel time annualpercentage        year
# Austria.1999          Austria 1999               NA          NA
# Belgium.1999          Belgium 1999               NA          NA
# Bulgaria.1999        Bulgaria 1999       -0.6806495          NA
# Estonia.1999          Estonia 1999               NA          NA
# France.1999            France 1999               NA          NA
# Germany.1999          Germany 1999               NA          NA
# Switzerland.1999  Switzerland 1999       -1.8497570          NA
# Austria.2000          Austria 2000       -0.6033900  0.14970015
# Belgium.2000          Belgium 2000               NA -0.49201756
# Bulgaria.2000        Bulgaria 2000        0.8263925 -0.36320990
# Estonia.2000          Estonia 2000               NA -2.51032544
# France.2000            France 2000               NA  0.57800624
# Germany.2000          Germany 2000               NA -0.52295712
# Switzerland.2000  Switzerland 2000        0.2783076  0.25616728
# Austria.2001          Austria 2001       -2.6962484 -0.15375642
# Belgium.2001          Belgium 2001        1.3088577  0.72528621
# Bulgaria.2001        Bulgaria 2001               NA          NA
# Estonia.2001          Estonia 2001               NA -0.05563662
# France.2001            France 2001        0.2224629  0.74205086
# Germany.2001          Germany 2001               NA -0.01185349
# Switzerland.2001  Switzerland 2001        0.8354322 -1.40826638
# Austria.2002          Austria 2002               NA          NA
# Belgium.2002          Belgium 2002               NA  1.60874778
# Bulgaria.2002        Bulgaria 2002               NA          NA
# Estonia.2002          Estonia 2002               NA  0.55866704
# France.2002            France 2002               NA -1.59866472
# Germany.2002          Germany 2002               NA -0.11217415
# Switzerland.2002  Switzerland 2002               NA          NA

数据

d <- structure(list(countrylabel = c("Austria", "Belgium", "Bulgaria", 
"Estonia", "France", "Germany", "Switzerland"), annualpercentage.1999 = c(NA, 
-2.58060150400384, -0.0623757258909573, 0.267776001395166, NA, 
NA, 0.048219924249952), annualpercentage.2000 = c(NA, -0.249416955035044, 
1.3525450891501, 1.04446768824697, NA, -0.0582347596434839, -0.891400228849837
), annualpercentage.2001 = c(1.82469277697851, NA, NA, 1.04231605324821, 
NA, -0.900145118946308, -1.19320727433597), year2000 = c(0.633712375393134, 
NA, 1.24760861316098, -0.092964787061478, -0.59403260962332, 
NA, -0.650348234181285), year2001 = c(0.587318286831079, NA, 
NA, 0.348890470222513, NA, NA, NA), year2002 = c(0.0645316087966406, 
-0.279456557428068, NA, NA, -0.0627400036074545, 1.30419117694731, 
-0.484654596062051)), row.names = c(NA, -7L), class = "data.frame")

Answer 3

回到原来的reshape代码，错误在于使用

list(2:ncol(d))

，它只创建了列数的列表：

list(2:ncol(d))

您需要的是带有列名称的向量：

list(colnames(d)[2:ncol(d)])

因此，

merge_data2 <- reshape(merge_data2, varying = list(colnames(merge_data2)[2:ncol(merge_data2)]), 
                       v.names = c("percentageshare", "Year"),
                       idvar = "countrylabel", direction = "long", times = 1990:2013)

将呈现您想要的内容。