Question

我试图通过引用列名从R中的data.table中提取特定值

require(data.table)

# Create data.frame

cohort = c("cohort1", "cohort2", "cohort3")
year = c(2019, 2018, 2020)
item_2018 = c("alpha", "beta", "gamma")
item_2019 = c("banana", "apples", "oranges")
item_2020 = c("Tim", "Daniel","Simon")
desired_result = c("banana", "beta", "Simon")  # the values in this column I want to programatically grab from the relevant column before

cohorts <- data.frame(cohort,year, item_2018, item_2019, item_2020, desired_result)



setDT(cohorts) # turn the data.frame into a data.table

setkey(cohorts, year)  # setting the key for the data.table (not sure if this is necessary)


# CALCULATE NEW FIELD (attempts - not working)

# trying to populate new column "result_attempt_1" with : c("banana", "beta", "Simon")

cohorts[, result_attempt_1 := get(paste0("item_", year)), by = year] # this returns c("Simon", "Simon", "Simon") rather than  c("banana", "beta", "Simon") 

cohorts[, result_attempt_2 := .SD[, get(paste0("item_", year)), by = year]] # very wrong

cohorts[, result_attempt_3 := .SD[, get(paste0("item_", year)), by = get(paste0("item_", year))]] # very wrong

我希望“ desired_results”列中的值最终出现在“ result_attempt”列中。我得到的最接近的是针对每个记录/行重复的最后一个正确结果。

任何想法如何实现这一目标？非常感谢。

Answer 1

一个选项是行/列索引

cohorts[, result := as.data.frame(.SD)[cbind(seq_len(.N),
      match(year, sub("item_", "", names(.SD))))], .SDcols = 3:5]

cohorts
#    cohort year item_2018 item_2019 item_2020 desired_result result
#1: cohort2 2018      beta    apples    Daniel           beta   beta
#2: cohort1 2019     alpha    banana       Tim         banana banana
#3: cohort3 2020     gamma   oranges     Simon          Simon  Simon

Answer 2

这是 a ：

for (rw in seq_len(nrow(cohorts))) {
  set(
    cohorts, 
    i = rw, 
    j = "newcol", 
    value = cohorts[[paste0("item_", cohorts[["year"]][rw])]][rw]
  )
}

> cohorts
    cohort year item_2018 item_2019 item_2020 desired_result newcol
1: cohort2 2018      beta    apples    Daniel           beta   beta
2: cohort1 2019     alpha    banana       Tim         banana banana
3: cohort3 2020     gamma   oranges     Simon          Simon  Simon

Answer 3

另一个使用data.table::melt的选项，匹配item_year然后查找并通过引用进行更新：

cohorts[
    melt(cohorts, measure.vars=patterns("^item"), variable.factor=FALSE)[, 
        value[variable==paste0("item_", year)], by=.(cohort)],
    on=.(cohort), desired_result := V1]

Answer 4

非常感谢您的有用回复。

似乎我面临的基本问题是data.table中列的数据类型。

似乎“ item_xxxx”列被强制为“因素”而不是“字符”。

如果我们从头开始定义data.table（而不是通过data.frame阶段），那么我最初提供的代码确实可以正常工作。

require(data.table)

# Create data.table

cohort = c("cohort1", "cohort2", "cohort3")
year = c(2019, 2018, 2020)
item_2018 = c("alpha", "beta", "gamma")
item_2019 = c("banana", "apples", "oranges")
item_2020 = c("Tim", "Daniel","Simon")
desired_result = c("banana", "beta", "Simon")  # the values in this column I want to programatically grab from the relevant column before


# create DATA.TABLE (not data.frame)

cohorts <- data.table(cohort,year, item_2018, item_2019, item_2020, desired_result)

str(cohorts)


# setDT(cohorts)

# trying to populate new column "result_attempt_1" with : c("banana", "beta", "Simon")

cohorts[, result_attempt_1 := get(paste0("item_", year)), by = year] # this now returns c("banana", "beta", "Simon"), as desired

cohorts

所以-在这里带回家消息-如果不起作用，请运行str（data_name）检查列数据类型。

感谢大家的帮助和支持。

R data.table如何基于另一列的值从多个列之一（按列NAME）获取VALUE

4 个答案: