dplyr fails to summarize - r

Question

My data frame is a time series with different variables and looks like
Date prcpmm TmaxC TminC
1/1/1966 0 7.8 0
1/2/1966 0.8 8.3 -1.1
1/3/1966 0 2.8 -5
1/4/1966 0 4.4 -5.6
1/5/1966 0 5.6 -8.9
1/6/1966 0 2.2 -1.7
1/7/1966 0.5 0.6 -12.8
1/8/1966 0 -6.1 -15
1/9/1966 0 3.9 -8.3
1/10/1966 0 5 -8.9
1/11/1966 0 -5 -12.2
1/12/1966 1 -3.3 -11.1
1/13/1966 1.3 2.2 -6.1
starting from 1966 - 2005, link to have an access to the input file: https://drive.google.com/open?id=0B09Gj4L5FPQWeGt0Wl94OWVlelE
I tried the blow mentioned code:
Adrian <- read.csv("~/Adrian.csv")
Adrian <- cbind(Adrian, weatherGen = rep("ObservedAdrian", nrow(Adrian)))
dateObj <- as.POSIXct(Adrian$Date, format = "%m/%d/%Y")
Adrian$Date <- dateObj
Adrian$day <- day(dateObj)
Adrian$month <- month(dateObj)
Adrian$year <- year(dateObj)
Adrian$prcpmm[Adrian$prcpmm==-99.00]<- NA
Adrian$TmaxC[Adrian$TmaxC==-99.00]<- NA
Adrian$TminC[Adrian$TminC==-99.00]<- NA
for (i in 1:length(Adrian$month)) {
if (Adrian$month[i] %in% 3:5) {
Adrian$season[i] = "spring"
Adrian$yearNew[i] = Adrian$year[i]
} else if (Adrian$month[i] %in% 6:8) {
Adrian$season[i] = "summer"
Adrian$yearNew[i] = Adrian$year[i]
} else if (Adrian$month[i] %in% 9:11) {
Adrian$season[i] = "fall"
Adrian$yearNew[i] = Adrian$year[i]
} else{
Adrian$season[i] = "winter"
Adrian$yearNew[i] = Adrian$year[i] + 1
}
}
Adrian$season <- ordered(Adrian$season, levels = c("winter", "spring",
"summer", "fall"))
It executed till this part very well without error.
Observed_everyseason_pVal <- Adrian %>% group_by(yearNew, season) %>%
summarise(pvalMK = mk.test(ts(prcpmm))$pvalue[1], SMK =
mk.test(ts(prcpmm))$Sg[1] )
Observed_everyseason_pVal1 <- Adrian %>% group_by(season) %>%
summarise(pvalMK = mk.test(ts(prcpmm))$pvalue[1], SMK =
mk.test(ts(prcpmm))$Sg[1])
from the above to code lines:
Observed_everyseason_pVal <- Adrian %>% group_by(yearNew, season)
It worked well without error but when I run with
Observed_everyseason_pVal <- Adrian %>% group_by(yearNew, season) %>%
summarise(pvalMK = mk.test(ts(prcpmm))$pvalue[1], SMK =
mk.test(ts(prcpmm))$Sg[1] )
It produces error:
Error in summarise_impl(.data, dots) : Column pvalMK is of
unsupported type NULL

Answer 1

if you do str(mk.test(ts(Adrian$prcpmm))), it gives you the following output:
List of 9
$ data.name : chr "ts(Adrian$prcpmm)"
$ p.value : num 0.259
$ statistic : Named num 1.13
..- attr(*, "names")= chr "z"
$ null.value : Named num 0
..- attr(*, "names")= chr "S"
$ parameter : Named int 13
..- attr(*, "names")= chr "n"
$ estimates : Named num [1:3] 16 176.67 0.28
..- attr(*, "names")= chr [1:3] "S" "varS" "tau"
$ alternative: chr "two.sided"
$ method : chr "Mann-Kendall trend test"
$ pvalg : num 0.259
- attr(*, "class")= chr "htest"
Therefore, the former object you specified, namely mk.test(ts(prcpmm))$pvalue[1], doesn´t exist. There is only mk.test(ts(prcpmm))$p.value[1]. Also, the object mk.test(ts(prcpmm))$Sg[1] (...$Sg) is not in scope at all. This explains the error: Error in summarise_impl(.data, dots) : Column pvalMK is of unsupported type NULL.