WeatherCompare.R

library(data.table)
check <- fread("https://storage.googleapis.com/playground-mike/weather/1315015.csv")
check[, .N, keyby=.(NAME)]

check <- check[, .(NAME, DATE,TMAX,TMIN)]
check <- na.omit(check, invert=FALSE)
check[, DATE := as.Date(DATE)]
check[, YEAR := year(DATE)]
check[, MONTH := month(DATE)]

# Check month count
check[, MonthCnt := .N, keyby=.(NAME, YEAR, MONTH)]
check <- check[MonthCnt > 25 & YEAR > 2011 & YEAR < 2018]

dcast(check[, .(Below32=sum(TMIN <= 32)), keyby=.(NAME, MONTH, YEAR)], YEAR+MONTH~NAME)
dcast(check[, .(Below32=sum(TMAX >= 95)), keyby=.(NAME, MONTH, YEAR)], YEAR+MONTH~NAME)

HighCompare <- dcast(check[, .(TMAX), keyby=.(NAME,DATE)], DATE~NAME)
HighCompare[month(DATE) %in% c(6,7,8), lapply(.SD, mean), .SDcols=-c("DATE")]
HighCompare[month(DATE) %in% c(12,1,2), lapply(.SD, mean), .SDcols=-c("DATE")]