R语言：基于DataFrame的tota_score列处理expected_score列数据

需求回顾 #

• 过滤掉tota_score列值为NA的行
• 仅保留expected_score中**大于对应行tota_score**的值，其余情况（包括小于等于、本身为NA）都设为NA

现有代码的问题 #

你写的ifelse(!is.na(expected_score) > !is.na(tota_score), NA, expected_score)逻辑完全错误——!is.na()返回布尔值（TRUE/FALSE），比较布尔值时会自动转成1和0，这根本不是在判断expected_score和tota_score的数值大小，自然得不到预期结果。

正确实现代码 #

使用dplyr包按顺序完成两个需求：

library(dplyr)

# 初始数据框
df <- data.frame(tota_score=c(4.5,12.2,4.6,9.2,12.2,NA,4.5,12.2,4.6,9.2,12.2,36.4),
                 expected_score=c(4.5,12.1,NA,10,12.2,36.4,5,12.5,NA,9.2,16,NA),
                 Region1=c("All region",NA,NA,"All region","All region",NA,"All region",NA,NA,"All region","All region",NA),
                 Region2=c("EAST","EAST","EAST","EAST","EAST",NA,"EAST","EAST","EAST","EAST","EAST",NA),
                 Region3=c("West",NA,"West","West","West","West","West",NA,"West","West","West","West"))

# 数据处理
df_processed <- df %>%
  # 过滤tota_score为NA的行
  filter(!is.na(tota_score)) %>%
  # 仅保留expected_score大于tota_score的值，其余设为NA
  mutate(expected_score = ifelse(expected_score > tota_score, expected_score, NA))

# 查看处理结果
print(df_processed)

代码解释 #

1. filter(!is.na(tota_score))：直接筛除tota_score为NA的行，完成第一个需求；
2. mutate(...)中的ifelse判断：
   • 当expected_score数值大于对应行的tota_score时，保留原数值；
   • 所有不满足条件的情况（包括expected_score小于等于tota_score、expected_score本身为NA），统一设为NA，完全匹配第二个需求。

处理后结果 #

tota_score expected_score     Region1 Region2 Region3
1         4.5             NA All region    EAST    West
2        12.2             NA        <NA>    EAST    <NA>
3         4.6             NA        <NA>    EAST    West
4         9.2            10.0 All region    EAST    West
5        12.2             NA All region    EAST    West
6         4.5             5.0 All region    EAST    West
7        12.2            12.5        <NA>    EAST    <NA>
8         4.6             NA        <NA>    EAST    West
9         9.2             NA All region    EAST    West
10       12.2            16.0 All region    EAST    West
11       36.4             NA        <NA>    <NA>    West

内容的提问来源于stack exchange，提问作者potro