R 用于在 1000 次迭代后生成 NA 的循环

Question

I have a simple for loop that I'm using to remove any rows from my dataframe that involve two variables sharing a similar string and when I run the loop it iterates 1000 times and then starts generating NA's which breaks my loop.

expiration	quote_datetime
2021-02-26	2021-02-26 10:00:00
2021-02-26	2021-02-27 10:00:00

for(row in 1:nrow(df)){
  if(grepl(df$expiration[row], df$quote_datetime[row],fixed=TRUE) == TRUE){
    df = df[-row,]
  }
}

I'm getting the error message

Error in if (grepl(df$expiration[row], df$quote_datetime[row], : missing value where TRUE/FALSE needed

Each time I run it it eliminates a few more rows until it runs out of anything else to eliminate and then it runs without error. Appreciate help.

Answer 1

The issue arises because the original data 'df' gets subset if the condition is TRUE, ie it will be one row less for every if TRUE case. It could be resolved if we copy of the data

df2 <- df
for(row in 1:nrow(df)){
   if(grepl(df$expiration[row], df$quote_datetime[row],fixed=TRUE)){
     df2 <- df2[-row,]
    }
   }

---

Also, grepl is vectorized only for the 'x' and not for the pattern So, if we need to do a vectorization, may need to paste the pattern together

df <- df[!grepl(paste(df$expiration, collapse="|"), 
              df$quote_datetime, fixed = TRUE), ]

---

Or use a function that does the vectorization for both 'x' and 'pattern ie str_detect

library(dplyr)
library(stringr)
df %>%
   filter(!str_detect(quote_datetime, fixed(expiration))