R语言中将字符型year转为因子时出现NA问题的排查与解决

我有如下数据集：

years <- c("2010", "2011", "2012", "2013", "2014", "2015", "2016", "2017", 
"2018", "2019", "2020")

n_cohorts <- length(years)

df <- structure(list(label2plot = structure(c(1L, 2L, 1L, 2L, 1L, 2L, 
1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L
), levels = c("aaa", "bbb"), class = c("ordered", 
"factor")), var_abs = c(717L, 569L, 860L, 752L, 713L, 575L, 918L, 
724L, 946L, 764L, 951L, 764L, 784L, 691L, 672L, 610L, 833L, 671L, 
773L, 620L, 532L, 293L), var_rel = c(0.557542768273717, 0.442457231726283, 
0.533498759305211, 0.466501240694789, 0.553571428571429, 0.446428571428571, 
0.559074299634592, 0.440925700365408, 0.553216374269006, 0.446783625730994, 
0.554518950437318, 0.445481049562682, 0.531525423728814, 0.468474576271186, 
0.52418096723869, 0.47581903276131, 0.553856382978723, 0.446143617021277, 
0.554917444364681, 0.445082555635319, 0.644848484848485, 0.355151515151515
), year = c("2020", "2020", "2019", "2019", "2018", "2018", 
"2017", "2017", "2016", "2016", "2015", "2015", "2014", "2014", 
"2013", "2013", "2012", "2012", "2011", "2011", "2010", "2010"
)), row.names = c(NA, -22L), class = "data.frame")

cohorts <- length(years)

我希望将其中的year列转换为因子，执行了如下代码：

df$year2  <- factor(df$year,
                    levels = c(1:n_cohorts),
                    labels = years)

但得到的结果中year2列全部为NA，无法解释原因，请问我哪里操作错误？结果如下：

structure(list(label2plot = structure(c(1L, 2L, 1L, 2L, 1L, 2L, 
1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L
), levels = c("aaa", "bbb"), class = c("ordered", "factor")), 
    var_abs = c(717L, 569L, 860L, 752L, 713L, 575L, 918L, 724L, 
    946L, 764L, 951L, 764L, 784L, 691L, 672L, 610L, 833L, 671L, 
    773L, 620L, 532L, 293L), var_rel = c(0.557542768273717, 0.442457231726283, 
    0.533498759305211, 0.466501240694789, 0.553571428571429, 
    0.446428571428571, 0.559074299634592, 0.440925700365408, 
    0.553216374269006, 0.446783625730994, 0.554518950437318, 
    0.445481049562682, 0.531525423728814, 0.468474576271186, 
    0.52418096723869, 0.47581903276131, 0.553856382978723, 0.446143617021277, 
    0.554917444364681, 0.445082555635319, 0.644848484848485, 
    0.355151515151515), year = c("2020", "2020", "2019", "2019", 
    "2018", "2018", "2017", "2017", "2016", "2016", "2015", "2015", 
    "2014", "2014", "2013", "2013", "2012", "2012", "2011", "2011", 
    "2010", "2010"), year2 = structure(c(NA_integer_, NA_integer_, 
    NA_integer_, NA_integer_, NA_integer_, NA_integer_, NA_integer_, 
    NA_integer_, NA_integer_, NA_integer_, NA_integer_, NA_integer_, 
    NA_integer_, NA_integer_, NA_integer_, NA_integer_, NA_integer_, 
    NA_integer_, NA_integer_, NA_integer_, NA_integer_, NA_integer_
    ), levels = c("2010", "2011", "2012", "2013", "2014", "2015", 
    "2016", "2017", "2018", "2019", "2020"), class = "factor")), row.names = c(NA, 
-22L), class = "data.frame")

错误原因 #

你在factor()函数中设置的levels参数是c(1:n_cohorts)，也就是数字序列1到11，但原始的year列是字符串类型的年份值（比如"2020"、"2019"等），两者的取值完全不匹配，R无法找到对应关系，因此所有值都被转换为NA。

正确解决方案 #

只需要将levels参数设置为years即可，因为years里的元素就是year列的所有取值，同时还能指定你想要的因子顺序（2010到2020）：

df$year2 <- factor(df$year, levels = years)

执行这段代码后，year2列会正确对应原始year的取值，并且因子的顺序按照years的顺序排列，不会出现NA。

内容的提问来源于stack exchange，提问作者Dierforth