CODEWITHSUNDEEP
rdplyr
roboes
roboes

Reputation: 401

Filter non-NAs on numeric column name

I am trying to filter a column which contains a numeric/date name using a as.Date variable.

As an example, consider this small database:

dt <- data.table(
names = c("A", "B", "C"),
`2020-01-01` = c(1, NA, 2),
`2020-01-02` = c(3, 4, 5),
`2020-01-03` = c(6, 7, 8)
)

I am currently filtering the desired date column as follows:

dt1 <- dt %>% filter(!is.na(`2020-01-01`)) %>% select(names)

My idea is to create a meeting_date variable, this variable should be used as a date reference for all my R code.

meeting_date <- as.Date("2020-01-01")

But of course the code:

dt1 <- dt %>% filter(!is.na(meeting_date)) %>% select(names)

Does not work. The reason for this is the missing backticks, so without success I tried the following codes:

dt1 <- dt %>% filter(!is.na(paste("`", meeting_date, "`", sep=""))) %>% select(names)
dt1 <- dt %>% filter(!is.na(noquote(paste("`", meeting_date, "`", sep="")))) %>% select(names)

Does anyone knows how to proceed? Thanks!

Upvotes: 1

Views: 298

Answers (3)

ThomasIsCoding
ThomasIsCoding

Reputation: 101753

You can use subset + is.naas below

meeting_date <- "2020-01-01"
dtout <- subset(dt,as.vector(!is.na(dt[, ..meeting_date])))

such that

> dtout
   names 2020-01-01 2020-01-02 2020-01-03
1:     A          1          3          6
2:     C          2          5          8

Upvotes: 0

mrhellmann
mrhellmann

Reputation: 5499

Long data should be easier to work with:

library(data.table)
dt <- data.table(
  names = c("A", "B", "C"),
  `2020-01-01` = c(1, NA, 2),
  `2020-01-02` = c(3, 4, 5),
  `2020-01-03` = c(6, 7, 8)
)

#Make data 'long' & change the new 'name' column to dates
# change confusing column 'name' to date while we're at it.
dt_long <- dt %>% pivot_longer(-names) %>% 
  mutate(date = lubridate::ymd(name)) %>%
  select(-name)

meeting_date <- as.Date("2020-01-01")

dt_long %>% filter(date == meeting_date & (!is.na(value)))

Upvotes: 0

tmfmnk
tmfmnk

Reputation: 39858

You can do:

meeting_date <- as.Date("2020-01-01")

dt %>%
 filter_at(vars(one_of(as.character(meeting_date))), ~ !is.na(.))

  names 2020-01-01 2020-01-02 2020-01-03
1     A          1          3          6
2     C          2          5          8

Upvotes: 2

Related Questions