Subsetting Data in R tool gives warning of incompatible methods

[Ashish]

Hello,

I am trying to subset my data based on a date, in the R tool and I get this warning

Warning: R (34): Incompatible methods ("Ops.factor", "Ops.POSIXt") for ">" 

 which  leads to undesirable results.

Pasted below is the code I am using (I have tried couple different lines of code but all of them gives the same warning).

All the options I tried (commented in the code below works fine in Rstudio)

# Read the data stream into R
Input <- read.Alteryx("#1", mode="data.frame")


season_matrix <- model.matrix(Input$Change ~Input$Season)[,-1]
Input <- cbind(Input,season_matrix)
#Input$Date <- as.Date(Input$Date)
#the.data_test <- Input[ Input$Date >= as.POSIXct("2015-12-31"),]
#the.data_train <- Input[ Input$Date <= as.POSIXct("2015-12-31"),]

the.data_test <- Input[ which(Input$Date > as.POSIXct("2015-12-31")),]
the.data_train <- Input[ which(Input$Date <= as.POSIXct("2015-12-31")),]

#the.data_test <- Input[ format(Input$Date,"%Y-%m-%d") >= "2015-12-31",]
#the.data_train <- Input[ format(Input$Date,"%Y-%m-%d") <= "2015-12-31",]

#the.data_test <- subset(Input , as.Date(Date) > as.POSIXct("2015-12-31"))
#the.data_train <- subset(Input , as.Date(Date) <= as.POSIXct("2015-12-31"))

Can anyone please give some idea on how to fix this.

Thanks,

Ashish

[Ashish · Accepted solution]

ok, I figured out the fix for the problem. Although I do not fully understand why it was not working earlier and why its working now, but somehow it worked.

I used as.Date function on both sides of operator to coerce the type.

require(glmnet)
require(stats)
# Read the data stream into R
Input <- read.Alteryx("#1", mode="data.frame")
season_matrix <- model.matrix(Input$Change ~Input$Season)[,-1]
Input <- cbind(Input,season_matrix)

#Input$Date <- as.Date(Input$Date)
#the.data_test <- Input[ Input$Date >= as.POSIXct("2015-12-31"),]
#the.data_train <- Input[ Input$Date <= as.POSIXct("2015-12-31"),]

the.data_test <- Input[ as.Date(Input$Date) >= as.Date("2015-12-31"),]
the.data_train <- Input[ as.Date(Input$Date) <= as.Date("2015-12-31"),]

The "Date" variable in the dataset was already of type date so I did not suspect it earlier.

Also when I tried formula tool create a new Date variable date format that did not work either.

I am relieved that I've found a fix but I lack the understanding what works and why.

Thanks,