如何在R中使用"within“时只过滤特定信息

Question

我试着在运行下面的代码后只过滤NA。但是我不能制定一个同样的代码。下面的代码在一个函数中，其中Raw_data最终根据公司划分为不同的数据帧。

Raw_data<-within(Raw_data,{
      Neg_outliers=NA
      Pos_outliers=NA
      Neg_outliers[Company == "2002"]<-ifelse(Paid_age<mean_two_sd_negative,"negative_outlier","NA")
      Neg_outliers[Company == "2203"]<-ifelse(Paid_age<mean_two_sd_negative,"negative_outlier","NA")
      Neg_outliers[Company == "1804"]<-ifelse(Paid_age<mean_two_sd_negative,"negative_outlier","NA")
      Neg_outliers[Company == "2401A"]<-ifelse(Paid_age<mean_two_sd_negative,"negative_outlier","NA")
      Neg_outliers[Company == "2401B"]<-ifelse(Paid_age<mean_two_sd_negative,"negative_outlier","NA")
      Pos_outliers[Company == "2002"]<-ifelse(Paid_age>mean_two_sd_positive,"positive_outlier","NA")
      Pos_outliers[Company == "2203"]<-ifelse(Paid_age>mean_two_sd_positive,"positive_outlier","NA")
      Pos_outliers[Company == "1804"]<-ifelse(Paid_age>mean_two_sd_positive,"positive_outlier","NA")
      Pos_outliers[Company == "2401A"]<-ifelse(Paid_age>mean_two_sd_positive,"positive_outlier","NA")
      Pos_outliers[Company == "2401B"]<-ifelse(Paid_age>mean_two_sd_positive,"positive_outlier","NA")

    })

Answer 1

这里不需要ifelse语句，因为新列中已经有了NA。正如@jay.sf指出的那样，在任何情况下，您都会将字符串"NA"而不是NA写入新列，这可能不是您想要的。

与其使用字符串或NA来指示值是否是异常值，为什么不使用TRUE和FALSE组成的逻辑列呢？这将使以后更容易过滤出您的数据。

您的整个代码可以替换为以下代码，它更易于阅读、理解和维护：

companies <- c("2202", "2203", "1804", "2401A", "2401B")
matching  <- Raw_data$Company %in% companies

Raw_data  <- within(Raw_data, {
      Pos_outliers <- matching & Paid_age > mean_two_sd_positive
      Neg_outliers <- matching & Paid_age < mean_two_sd_positive
})

因此，如果要添加新公司，只需向companies添加单个值，而不是一整行代码；如果要更改正在测试的变量，只需更改一个或两个变量名，而不是十个。