r - 根据查找表（多列）在数据框中查找（和替换）值

Question

下面更新

：原来的

我试图找到基于匹配来自另一个数据帧的两列的最优雅（简单和简洁）的方法来替换某些列的值。

这是包含我希望替换的列的表（基于它们包含的值）。

> cost.table
  Identifier Phase.0.Difficulty Phase.1.Complexity Phase.2.Complexity Phase.3.Complexity Phase.4.Complexity Phase.5.Complexity
1        FS1                Low                Low                Low             Medium             Medium               High
2        FS2               High               High               High             Medium             Medium             Medium
3        FS3               High                Low                Low               High               High               High
4        FS4               High             Medium             Medium             Medium             Medium             Medium
5        FS5               High             Medium             Medium               High             Medium             Medium
  Phase.6.Complexity Transaction.Feasibility Approach
1               High                  Medium        B
2             Medium                  Medium        I
3               High                  Medium        B
4             Medium                  Medium        I
5             Medium                  Medium        B

以下是我希望用来查找正确替换值的查找表。

> cost.approach.difficulty
  Approach Difficulty   Phase 0  Phase 1  Phase 2   Phase 3  Phase 4  Phase 5  Phase 6
1        B       High 18102.778 29481.67 29481.67 11822.222 30737.78 21634.67 12768.00
2        B        Low  3860.694 15978.47 11175.69  7448.000 12768.00 11467.56 11467.56
3        B     Medium  5323.694 24974.44 15184.17  9221.333 15368.89 12768.00 12768.00
4        I       High 18102.778 74184.44 29481.67 44747.111 69160.00 45249.56 32245.11
5        I        Low  3860.694 26008.89 11175.69 16551.111 35910.00 16876.22 14275.33
6        I     Medium  5323.694 41156.11 15184.17 22373.556 44776.67 23378.44 16876.22
7       RV       High 18102.778 28373.33 29481.67 44747.111 69160.00 45249.56 32245.11
8       RV        Low  3860.694 14870.14 11175.69 16551.111 44776.67 16876.22 14275.33
9       RV     Medium  5323.694 22757.78 15184.17 22373.556 44776.67 23378.44 16876.22

我试图找到一个简单的解决方案来在 cost.approach.difficulty 表中查找“方法”和“难度”的相应值。

因此，例如，在 cost.table 中，我想将第一行 Phase.0.Difficulty 列替换为 3860.694（因为它是“B”方法且难度较低。

有没有人有一个优雅、简单的解决方案来查找基于两列（或更多列）的值并沿多列替换值？

谢谢，

安德鲁

更新 -

有两个与使用合并相关的建议答案。我的目标是找到一个更简洁、简洁、优雅的解决方案。这是迄今为止我想出的最好的：

cost.approach.difficulty$Phase.0[match(paste(cost.table$Approach, cost.table$Phase.0.Difficulty), paste(cost.approach.difficulty$Approach, cost.approach.difficulty$Difficulty))]

这个解决方案的问题是我需要提前知道列名，而且看起来仍然像一个黑客。有人有更简洁的解决方案吗？

Answer 1

如果您希望这适用于可变数量的列，我建议将您的成本表和查找表重塑为更标准化的格式。

首先，如果您以可重现的格式提供数据，那么回答这个问题会更容易：

# Create the example data
cost.table <- data.frame(
  "Identifier" = c("FS1", "FS2",  "FS3",  "FS4",  "FS5"),
  "Phase.0.Difficulty" = c("Low", "High", "High", "High", "High"),
  "Phase.1.Complexity" = c("Low", "High", "Low", "Medium", "Medium"),
  "Phase.2.Complexity" = c("Low", "High", "Low", "Medium", "Medium"),
  "Phase.3.Complexity" = c("Medium", "Medium", "High", "Medium", "High"),
  "Phase.4.Complexity" = c("Medium", "Medium", "High", "Medium", "Medium"),
  "Phase.5.Complexity" = c("High", "Medium", "High", "Medium", "Medium"),
  "Phase.6.Complexity" = c("High", "Medium", "High", "Medium", "Medium"),
  "Transaction.Feasibility" = c("Medium", "Medium", "Medium", "Medium", "Medium"),
  "Approach" = c("B", "I", "B", "I", "B"),
  stringsAsFactors = FALSE)

cost.approach.difficulty <- data.frame(
  "Approach" = c("B", "B", "B", "I", "I", "I", "RV", "RV", "RV"),
  "Difficulty" = c("High", "Low", "Medium", "High", "Low", "Medium", "High", "Low", "Medium"),
  "Phase.0" = c(18102.778, 3860.694, 5323.694, 18102.778, 3860.694, 5323.694, 18102.778, 3860.694, 5323.694),
  "Phase.1" = c(29481.67,15978.47, 24974.44, 74184.44, 26008.89, 41156.11, 28373.33, 14870.14, 22757.78),
  "Phase.2" = c(29481.67, 11175.69, 15184.17, 29481.67, 11175.69, 15184.17, 29481.67, 11175.69, 15184.17),
  "Phase.3" = c(11822.222, 7448, 9221.333, 44747.111, 16551.111, 22373.556, 44747.111, 16551.111, 22373.556),
  "Phase.4" = c(30737.78, 12768, 15368.89, 69160, 35910, 44776.67, 69160, 44776.67, 44776.67),
  "Phase.5" = c(21634.67, 11467.56, 12768, 45249.56, 16876.22, 23378.44, 45249.56, 16876.22, 23378.44),
  "Phase.6" = c(12768, 11467.56, 12768, 32245.11, 14275.33, 16876.22, 32245.11, 14275.33, 16876.22),
  stringsAsFactors = FALSE)

一旦我重新创建了您的示例数据，我就使用melt.data.frame了包中的函数reshape2：

# Reshape the data
require(reshape2)

cost.table <- melt(cost.table, id.vars = c("Identifier", "Approach"), 
  value.name = "Size")
cost.table$Phase <- gsub("(\\w+\\.\\d+)\\.(\\w+)", "\\1", 
  as.character(cost.table$variable), perl = TRUE)
cost.table$Type <- gsub("(\\w+\\.\\d+)\\.(\\w+)", "\\2", 
  as.character(cost.table$variable), perl = TRUE)

head(cost.table)

  Identifier Approach           variable Size   Phase       Type
1        FS1        B Phase.0.Difficulty  Low Phase.0 Difficulty
2        FS2        I Phase.0.Difficulty High Phase.0 Difficulty
3        FS3        B Phase.0.Difficulty High Phase.0 Difficulty
4        FS4        I Phase.0.Difficulty High Phase.0 Difficulty
5        FS5        B Phase.0.Difficulty High Phase.0 Difficulty
6        FS1        B Phase.1.Complexity  Low Phase.1 Complexity

cost.approach.difficulty <- melt(cost.approach.difficulty, 
  id.vars = c("Difficulty", "Approach"), variable.name = "Phase")
cost.approach.difficulty$Phase <- as.character(cost.approach.difficulty$Phase)
cost.approach.difficulty$Type <- "Difficulty"
colnames(cost.approach.difficulty)[
  colnames(cost.approach.difficulty) == "Difficulty"] <- "Size"

head(cost.approach.difficulty)

    Size Approach   Phase     value       Type
1   High        B Phase.0 18102.778 Difficulty
2    Low        B Phase.0  3860.694 Difficulty
3 Medium        B Phase.0  5323.694 Difficulty
4   High        I Phase.0 18102.778 Difficulty
5    Low        I Phase.0  3860.694 Difficulty
6 Medium        I Phase.0  5323.694 Difficulty

一旦这两个表采用标准化格式，您就可以调用merge：

cost.table.filled <- merge(cost.table, cost.approach.difficulty, 
  by = c("Approach", "Size", "Phase", "Type"), all.x = TRUE, all.y = FALSE)

然后，如果您没有查找某些列的值，则可以重新插入原始值（否则您最终会得到一堆 NA）：

cost.table.filled$value[is.na(cost.table.filled$value)] <- 
  cost.table.filled$Size[is.na(cost.table.filled$value)]

然后你可以dcast把东西恢复成原来的格式：

cost.table.final <- dcast(cost.table.filled, Identifier + Approach ~ Phase + Type)

head(cost.table.final)

  Identifier Approach Phase.0_Difficulty Phase.1_Complexity Phase.2_Complexity Phase.3_Complexity Phase.4_Complexity Phase.5_Complexity Phase.6_Complexity Transaction.Feasibility_Transaction.Feasibility
1        FS1        B           3860.694                Low                Low             Medium             Medium               High               High                                          Medium
2        FS2        I          18102.778               High               High             Medium             Medium             Medium             Medium                                          Medium
3        FS3        B          18102.778                Low                Low               High               High               High               High                                          Medium
4        FS4        I          18102.778             Medium             Medium             Medium             Medium             Medium             Medium                                          Medium
5        FS5        B          18102.778             Medium             Medium               High             Medium             Medium             Medium                                          Medium

要替换所有列，我将melt每个查找表，然后将cbind它们全部放在一个查找表中。这样，您只需调用merge一次，而不必担心更换 NA。

Answer 2

在这种情况下，merge应该做的伎俩：

cost.table <- merge(
  x = cost.table,
  y = cost.approach.difficulty[c("Approach", "Difficulty", "Phase.0")],
  by.x = c("Phase.0.Difficulty", "Approach"),
  by.y = c("Difficulty", "Approach"), sort = FALSE
)
cost.table$Phase.0.Difficulty <- NULL
names(cost.table)[names(cost.table) == "Phase.0"] <- "Phase.0.Difficulty"

cost.table
  Approach Identifier Phase.1.Complexity Phase.2.Complexity Phase.3.Complexity Phase.4.Complexity Phase.5.Complexity Phase.6.Complexity Transaction.Feasibility Phase.0.Difficulty
1        B        FS1                Low                Low             Medium             Medium               High               High                  Medium           3860.694
2        I        FS2               High               High             Medium             Medium             Medium             Medium                  Medium          18102.778
3        I        FS4             Medium             Medium             Medium             Medium             Medium             Medium                  Medium          18102.778
4        B        FS3                Low                Low               High               High               High               High                  Medium          18102.778
5        B        FS5             Medium             Medium               High             Medium             Medium             Medium                  Medium          18102.778

Answer 3

最简单的答案似乎是：

• 使用粘贴组合查找列
• 使用 match 从查找表中查找行号

下面的代码用一行完成了多列查找。

    cost.approach.difficulty$Phase.0[match(paste(cost.table$Approach, 
cost.table$Phase.0.Difficulty), paste(cost.approach.difficulty$Approach, 
cost.approach.difficulty$Difficulty))]

要循环浏览多个列，for 循环可以正常工作。

不幸的是，我希望有一个本机解决方案，它可能采用列向量并将它们组合起来进行查找，但我还没有找到它。我将检查其他软件包以查看是否存在此类功能。