如何基于时间维度的区域合并/拆分对数据集R中的区域进行分组？

原始数据 #

data <- data.frame(
  id_territory = c("id_1", "id_2", "id_3","id_3","id_4","id_5","id_6","id_7", "id_8", "id_9"),
  type_mutation = c("absorbed territory","absorbed territory", "absorbing territory", "absorbed territory", "absorbing territory", "absorbing territory","creation","part of territory transferred", "absorbed territory", "absorbing territory"),
  id_reform = c("A","A","A","B","B","B","C","C","D","D")
)

变更说明 #

• 改革A：id_1、id_2被id_3合并
• 改革B：id_3拆分为id_4、id_5
• 改革C：id_6由id_7的部分区域新建
• 改革D：id_8被id_9合并

期望输出 #

df <- data.frame(
  id_territory = c("id_1", "id_2", "id_3","id_3","id_4","id_5","id_6","id_7", "id_8", "id_9"),
  type_mutation = c("absorbed territory","absorbed territory", "absorbing territory", "absorbed territory", "absorbing territory", "absorbing territory","creation","part of territory transferred", "absorbed territory", "absorbing territory"),
  id_reform = c("A","A","A","B","B","B","C","C","D","D"),
  group_id = c("group_1","group_1", "group_1", "group_1","group_1","group_1","group_2","group_2", "group_3", "group_3")
)

实现代码 #

核心思路是用连通分量识别关联区域：把每个改革中存在变更关系的区域连边，同一连通分量的区域归为同一组。

# 安装加载igraph包
if (!require(igraph)) {
  install.packages("igraph")
  library(igraph)
}

# 构建关联边列表
edges <- data.frame(from = character(), to = character())

# 遍历每个改革批次，生成关联边
for (reform in unique(data$id_reform)) {
  sub_data <- subset(data, id_reform == reform)
  
  # 处理吸收/合并场景：被吸收区域关联到吸收区域
  absorbing <- sub_data$id_territory[sub_data$type_mutation == "absorbing territory"]
  absorbed <- sub_data$id_territory[sub_data$type_mutation == "absorbed territory"]
  if (length(absorbing) > 0 && length(absorbed) > 0) {
    edges <- rbind(edges, data.frame(from = absorbed, to = rep(absorbing[1], length(absorbed))))
  }
  
  # 处理新建/部分转移场景：新建区域关联到转出区域
  creation <- sub_data$id_territory[sub_data$type_mutation == "creation"]
  transferred <- sub_data$id_territory[sub_data$type_mutation == "part of territory transferred"]
  if (length(creation) > 0 && length(transferred) > 0) {
    edges <- rbind(edges, data.frame(from = creation, to = transferred))
  }
}

# 构建无向图并提取连通分量
g <- graph_from_data_frame(edges, directed = FALSE)
comp <- components(g)

# 为每个区域分配分组ID
data$group_id <- paste0("group_", comp$membership[match(data$id_territory, names(comp$membership))])

# 处理无关联的独立区域（示例中无此情况，保留鲁棒性）
unlinked <- setdiff(unique(data$id_territory), names(comp$membership))
if (length(unlinked) > 0) {
  next_group <- max(comp$membership) + 1
  for (terr in unlinked) {
    data$group_id[data$id_territory == terr] <- paste0("group_", next_group)
    next_group <- next_group + 1
  }
}

# 输出结果
print(data)

运行后即可得到期望的分组结果：id_1-id_5因多次拆分合并属于同一连通分量（group_1），id_6-id_7因新建关联属于group_2，id_8-id_9因合并属于group_3。

内容的提问来源于stack exchange，提问作者Coralie