时间序列数据事件前后行提取及中心化时序图绘制技术问题

目标 #

从时间序列数据框df中，绘制特定状态（如状态A）每次出现时，其onset（起始行）前后指定时间点的图形。x轴以0为中心，事件前的时间点为负值，事件后的为正值，原理与**刺激时间直方图（peristimulus time histogram）**一致。

数据说明 #

我们有一组状态持续时长可变的时间序列数据，先通过游程编码（run length encoding）确定各状态的起止，再用以下函数提取状态A onset前1行、后2行的数据，得到extracted数据集：

# 数据构建
df <- data.frame(
  state =      c("A","A","A","A","A","B","A","A","X","Y","Z","A","A","A","B","A","A"),
  start =      c("start","NA","NA","NA","NA","NA","start","NA","NA","NA","NA","start","NA","NA","NA","start","NA"),
  rleGroup =   c("1","1","1","1","1","2","3","3","4","5","6","7","7","7","8","9","9"),
  data = runif(17)
) %>%
  mutate(original_row = row_number()) %>% # 添加原始行号用于后续匹配
  tidyr::unite(stateStart, c(state,start), sep = ".", remove = FALSE)

# 提取函数
extract.with.context <- function(x, colname, rows, after = 0, before = 0) {
  match.idx  <- which(x[[colname]] %in% rows)
  span       <- seq(from = -before, to = after)
  extend.idx <- c(outer(match.idx, span, `+`))
  extend.idx <- Filter(function(i) i > 0 & i <= nrow(x), extend.idx)
  extend.idx <- sort(unique(extend.idx))
  return(x[extend.idx, , drop = FALSE]) 
}
extracted = extract.with.context(x=df, colname="stateStart", rows=c("A.start"), after = 2, before = 1)

待解决问题 #

1. 创建唯一plotGroup标识：为每个包含A onset前1行、后2行的分组生成唯一标识，现有尝试未得到预期效果。
2. 创建中心化span计数器：生成以0为中心的计数器，标记A onset前1行（值为-1）、起始行（值为0）、后2行（值为1、2）。

解决方案 #

我们可以基于原数据中A onset的位置，为提取出的每条数据匹配对应的分组和span值：

library(dplyr)
library(purrr)

# 获取所有A onset在原数据中的位置
onset_indices <- which(df$stateStart == "A.start")
before_n <- 1
after_n <- 2

# 生成每个onset对应的分组和span映射表
group_span_map <- map_dfr(seq_along(onset_indices), function(group_id) {
  onset_row <- onset_indices[group_id]
  # 计算该onset覆盖的行范围
  target_rows <- seq(onset_row - before_n, onset_row + after_n)
  # 过滤掉超出数据框范围的无效行
  valid_rows <- target_rows[target_rows >= 1 & target_rows <= nrow(df)]
  # 生成对应的中心化span值
  span_values <- seq(-before_n, after_n)[seq_along(valid_rows)]
  
  tibble(
    original_row = valid_rows,
    plotGroup = group_id,
    span = span_values
  )
})

# 将分组和span信息合并到extracted数据集中
extracted <- extracted %>%
  left_join(group_span_map, by = "original_row")

关键说明 #

• 通过保留原始行号，确保提取后的数据能精准匹配到对应的onset分组
• 为每个onset生成独立的plotGroup编号，保证分组唯一性
• 按onset位置生成中心化的span值，直接对应相对时间点

最终绘图 #

使用ggplot绘制各分组的时序图：

library(ggplot2)
ggplot(data=extracted, aes(x=span, y = data, group = plotGroup)) + 
  geom_line() +
  labs(x = "相对于A onset的时间点", y = "数据值")

内容的提问来源于stack exchange，提问作者redatoms