Hello everyone, I am Deng Fei. I haven't updated my blog for a long time, because I haven't made progress for a long time.
I thought Lu Xun was right before. He wrote in "Wild Grass": "When I am silent, I feel full; when I speak, I feel empty at the same time." The exact situation now is that when I stop updating, I feel full and stress-free, then I don’t want to update more and more, and finally I find that there is nothing to write, and once I want to write something, I feel very empty, my stomach is empty but I started to have a big belly, as if there was nothing in my stomach, but it was all meat, the sadness of an adult...
Back on track, when I used dplyr's filter to control data today, unexpected results appeared. In line with the principle of "learning should be shared, and output is the best learning", I simulated a data and introduced this pit. and how to avoid it.
1. First simulate a set of data
set.seed(123)
df = data.frame(ID = 1:10, Sex = c("F","F","F","F","NA","F","F","NA","M","M"), y1 = c(rnorm(9),NA))
df
Data are as follows:
> df
ID Sex y1
1 1 F -0.56047565
2 2 F -0.23017749
3 3 F 1.55870831
4 4 F 0.07050839
5 5 NA 0.12928774
6 6 F 1.71506499
7 7 F 0.46091621
8 8 NA -1.26506123
9 9 M -0.68685285
10 10 M NA
2. Extract the line where Sex is not F
I have three methods:
- The first one, use !=
- The second, use ! ==
- The third way, use !%in%
The sample code is as follows:
df %>% filter(Sex != "F")
df %>% filter(!Sex == "F")
df %>% filter(!Sex %in% "F")
Example result:
It can be seen that the results of the three are consistent.
3. If the data is saved to Excel and then read
write.xlsx(df,"df_test.xlsx")
Read excel data:
df = read.xlsx("df_test.xlsx")
df
4. Weird moment: Excel reading error
library(tidyverse)
df %>% filter(Sex != "F")
df %>% filter(!Sex == "F")
df %>% filter(!Sex %in% "F")
The first two are wrong, it automatically ignores rows with NA...
Only the third is correct:
5. The data frame built in R is fine, but Excel is broken after a circle
It's that weird.
Complete code:
set.seed(123)
df = data.frame(ID = 1:10, Sex = c("F","F","F","F","NA","F","F","NA","M","M"), y1 = c(rnorm(9),NA))
df
#
library(tidyverse)
library(openxlsx)
df %>% filter(Sex != "F")
df %>% filter(!Sex == "F")
df %>% filter(!Sex %in% "F")
write.xlsx(df,"df_test.xlsx")
# 读取数据
df1 = read.xlsx("df_test.xlsx")
df1
str(df1)
library(tidyverse)
df1 %>% filter(Sex != "F")
df1 %>% filter(!Sex == "F")
df1 %>% filter(!Sex %in% "F")
Compare the two dataframes: R has NA, Excel reads <NA>
,
Use drop_na to see if it is a missing value:
It turns out that when I build the vector in R, I use "NA"
, instead of NA
, as characters, so filter != can extract NA rows.
6. Rebuild the R data frame
set.seed(123)
df2 = data.frame(ID = 1:10, Sex = c("F","F","F","F",NA,"F","F",NA,"M","M"), y1 = c(rnorm(9),NA))
df2
Try it with drop_na, no problem:
> df2 %>% drop_na(Sex)
ID Sex y1
1 1 F -0.4456620
2 2 F 1.2240818
3 3 F 0.3598138
4 4 F 0.4007715
5 6 F -0.5558411
6 7 F 1.7869131
7 9 M -1.9666172
8 10 M NA
Filter with three methods, try it: the first two are not ideal.
> df2 %>% filter(Sex != "F")
ID Sex y1
1 9 M -1.966617
2 10 M NA
> df2 %>% filter(!Sex == "F")
ID Sex y1
1 9 M -1.966617
2 10 M NA
> df2 %>% filter(!Sex %in% "F")
ID Sex y1
1 5 <NA> 0.1106827
2 8 <NA> 0.4978505
3 9 M -1.9666172
4 10 M NA
Conclusion: When filtering, the NA rows will be automatically ignored, so it is %in%
reliable to use it! ! !