将数据与 data.table roll="nearest" 合并在整个 DF 中滚动匹配答案

【问题标题】：Merging data with data.table roll="nearest" rolls matches across the entire DF将数据与 data.table roll="nearest" 合并在整个 DF 中滚动匹配
【发布时间】：2019-12-06 19:09:27
【问题描述】：

我有两个表，一个示例表和一个消息表。在消息表中，消息记录在跟踪器的采样率之外。我一直在做的是使用 data.table roll 最近将示例消息时间与示例报告中最接近的值相匹配。与其将示例消息返回到最近的时间并为其他所有内容返回 NA，似乎它正在滚动消息直到下一条消息。


library(data.table)
remotes::install.github("dmirman/gazer") # to get the data
library((gazer)

samp <- system.file("extdata", "TJ_samp.xls", package = "gazer")
samp <- data.table::fread(samp, stringsAsFactors = FALSE) # reads in large datasets 

msg <- system.file("extdata", "TJ_msg.xls", package = "gazer")
msg <- data.table::fread(msg, stringsAsFactors = FALSE) # reads in large datasets

setDT(samp)
setDT(msg)

DT_mesg <- msg[samp, on="time", roll="nearest"] # use this to get close to values in sample report

DT_mesg

 #SR edfs are a nightmare. This makes it so messages are alined with closest values 
get_msg <- DT_mesg %>% 
  group_by(trial, message) %>% 
  top_n(n=1, wt=desc(time)) # there are a lot of useless messages and they occupy the same time stamp. Only take the first message in time. This was one way I tried to deal with the multiple message issue, but it does not return messages close to their actual time. 

get_msg

【问题讨论】：

标签： r data.table

【解决方案1】：

我想通了……

setDT(samp)
setDT(msg)

DT_mesg <- msg[samp, on="time", roll=4]

这给了我想要的结果：


  trial    time                     message i.trial     x     y  pup Label
     1:     1 3314705 !MODE RECORD CR 250 2 1 L\n       1 958.8 580.8 4043  <NA>
     2:    NA 3314709                        <NA>       1 959.1 576.2 4052  <NA>
     3:    NA 3314713                        <NA>       1 959.8 575.6 4053  <NA>
     4:    NA 3314717                        <NA>       1 960.6 575.2 4056  <NA>
     5:    NA 3314721                        <NA>       1 960.2 579.6 4049  <NA>

不知道为什么 roll="nearest" 返回这个：

   trial    time                     message i.trial     x     y  pup Label
     1:     1 3314705 !MODE RECORD CR 250 2 1 L\n       1 958.8 580.8 4043  <NA>
     2:     1 3314709 !MODE RECORD CR 250 2 1 L\n       1 959.1 576.2 4052  <NA>
     3:     1 3314713 !MODE RECORD CR 250 2 1 L\n       1 959.8 575.6 4053  <NA>
     4:     1 3314717 !MODE RECORD CR 250 2 1 L\n       1 960.6 575.2 4056  <NA>
     5:     1 3314721 !MODE RECORD CR 250 2 1 L\n       1 960.2 579.6 4049  <NA>

【讨论】：