I am working with weatherAUS dataset that is avaiable in the R libraries. I am trying to replace "Yes" to 1 and "No" to 0 in the RainTomorrow column.
I wrote this but it doesn't seem to work:
weather4$RainTomorrow[weather4$RainTomorrow=="Yes"]<-1
I just says:
Warning message: In [<-.factor
(*tmp*
, weather4$RainTomorrow ==
"Yes", value = c(NA, : invalid factor level, NA generated
What does it mean and what should I do? I reckon I should use as.numeric or as.factor somewhere but I don't know how really.
You can easily do this with dplyr.
require(dplyr)
weather4 <- weather4 %>%
mutate(RainToday = ifelse(RainToday == "No",0,1))
Hope this helps
library(data.table)
weather4[,":="(RainTomorrow=ifelse(RainTomorrow=="no",0,1))]
or simply use:
as.numeric(as.factor(weather4$RainTomorrow))
This is a fairly common thing when one is testing different models. For instance, decision trees work well with "Yes" and "No". However some Regression models demands 1 and 0. Particular Logistic Regression.
I solved this by using the plyr library. It was extremely easy and convenient. Here is my solution.
Origin of solution is here.
library(plyr)
weather5$RainToday <- revalue(weather5$RainToday, c("Yes"=1))
weather5$RainToday <- revalue(weather5$RainToday, c("No"=0))
head(weather5$RainToday)
[1] 0 1 1 1 1 0
Levels: 0 1
Peace!