如何创建从其他字符串读取结果的字符列？

Question

我有过

       measure           estimate.difference
1   pT8 vs pT7  0.33 (95% CI: -1.39 to 2.06)
2 No pT vs pT7 -2.31 (95% CI: -4.95 to 0.33)
3 No pT vs pT8 -2.64 (95% CI: -5.67 to 0.38)
4   pT8 vs pT7  0.14 (95% CI: -0.26 to 0.53)
5 No pT vs pT7  0.16 (95% CI: -0.37 to 0.69)

df$estimate.difference比较了df$measure中列出的两个独立预测模型的结果。

如果df$estimate.difference为正，则... vs ...的“左侧”是最佳模型，如果值为负值，则“右侧”为最佳。我需要一个列来表示什么模型是最好的(我的数据集很大)。

是否可以创建一个字符列，该列首先读取df$estimate.difference是正的还是负的，然后打印左或右的“度量值名称”？

预期输出

       measure       best.model               estimate.difference
1   pT8 vs pT7              pT8       0.33 (95% CI: -1.39 to 2.06)
2 No pT vs pT7              pT7      -2.31 (95% CI: -4.95 to 0.33)
3 No pT vs pT8              pT8      -2.64 (95% CI: -5.67 to 0.38)
4   pT8 vs pT7              pT8       0.14 (95% CI: -0.26 to 0.53)
5 No pT vs pT7            No pT       0.16 (95% CI: -0.37 to 0.69)

数据

df <- structure(list(measure = c("pT8 vs pT7", "No pT vs pT7", "No pT vs pT8", 
                           "pT8 vs pT7", "No pT vs pT7"), estimate.difference = c("0.33 (95% CI: -1.39 to 2.06)", 
                                                                                  "-2.31 (95% CI: -4.95 to 0.33)", "-2.64 (95% CI: -5.67 to 0.38)", 
                                                                                  "0.14 (95% CI: -0.26 to 0.53)", "0.16 (95% CI: -0.37 to 0.69)"
                           )), row.names = c(NA, 5L), class = "data.frame")

Answer 1

使用strsplit和mapply。

transform(df, 
          best.model=mapply(`[`, strsplit(df$measure, ' vs '),
                            (as.numeric(sapply(strsplit(df$estimate.difference, ' '), `[`, 1)) < 0 ) + 1)
          )
#        measure           estimate.difference best.model
# 1   pT8 vs pT7  0.33 (95% CI: -1.39 to 2.06)        pT8
# 2 No pT vs pT7 -2.31 (95% CI: -4.95 to 0.33)        pT7
# 3 No pT vs pT8 -2.64 (95% CI: -5.67 to 0.38)        pT8
# 4   pT8 vs pT7  0.14 (95% CI: -0.26 to 0.53)        pT8
# 5 No pT vs pT7  0.16 (95% CI: -0.37 to 0.69)      No pT