拆分单列数据统计出版物数量与总引用数（R/Stata/Python）

R 实现 #

借助tidyverse工具集的stringr包即可完成处理：

library(tidyverse)

# 构造示例数据
df <- tibble(
  raw_data = c("21070808(136)|19995886(87)|21280165(66)", "20226255(57)|21440646(54)")
)

# 生成目标列
df_processed <- df %>%
  mutate(
    # 统计出版物数量：|的个数加1
    pub_count = str_count(raw_data, "\\|") + 1,
    # 提取所有括号内的引用数并求和
    total_citations = map_int(raw_data, ~{
      str_extract_all(.x, "\\((\\d+)\\)")[[1]] %>%
        str_remove_all("[()]") %>%
        as.integer() %>%
        sum()
    })
  ) %>%
  select(pub_count, total_citations)

# 查看结果
print(df_processed)

输出结果：

# A tibble: 2 × 2
  pub_count total_citations
      <int>           <int>
1         3             289
2         2             111

Stata 实现 #

通过正则表达式和循环处理，两种方式可选：

方式1：直接提取所有引用数 #

* 示例数据
clear
input strL raw_data
"21070808(136)|19995886(87)|21280165(66)"
"20226255(57)|21440646(54)"
end

* 计算出版物数量
gen pub_count = regexrcount(raw_data, "\|") + 1

* 提取并累加引用数
gen total_citations = 0
tempvar cit_str
forvalues i = 1/100 { // 假设每行最多100个出版物，可按需调整
  capture regexs(raw_data, "\((\d+)\)", `i', `cit_str')
  if _rc break
  replace total_citations = total_citations + real(`cit_str')
}

* 查看结果
list pub_count total_citations

方式2：拆分后逐个处理 #

split raw_data, parse("|")
gen pub_count = r(nvars)
gen total_citations = 0
forvalues i = 1/`r(nvars)' {
  replace total_citations = total_citations + real(ustrregexs(1)) if ustrregexm(raw_data`i', "\((\d+)\)")
}
drop raw_data1-raw_data`r(nvars)'

Python 实现 #

用pandas结合正则表达式快速处理：

import pandas as pd
import re

# 示例数据
raw_data = [
    "21070808(136)|19995886(87)|21280165(66)",
    "20226255(57)|21440646(54)"
]
df = pd.DataFrame({"raw_data": raw_data})

# 定义行处理函数
def calc_stats(row):
    entries = row.split("|")
    pub_count = len(entries)
    # 提取所有括号内的数字并求和
    citations = [int(re.search(r"\((\d+)\)", entry).group(1)) for entry in entries]
    return pd.Series([pub_count, sum(citations)])

# 生成目标列
df[["pub_count", "total_citations"]] = df["raw_data"].apply(calc_stats)

# 输出结果
print(df[["pub_count", "total_citations"]])

输出结果：

pub_count  total_citations
0          3              289
1          2              111

内容的提问来源于stack exchange，提问作者ClaraJ