如果开始月份不是1月份，请指定年份编号

Question

我有一个df data.frame，包含8年的每日价值。

date <- rep(as.Date(seq(as.Date("2001-05-01"), 
        as.Date("2008-04-30"), by= 1), format="%Y-%m-%d"), 3) 

site <- c(rep("Site_1", 2557), rep("Site_2", 2557), rep("Site_3", 2557)) 

value <- c(as.numeric(sample(90:271, 2557, replace=T)), 
      as.numeric(sample(125:340, 2557, replace=T)), 
      as.numeric(sample(70:173, 2557, replace=T))) 

df <- data.frame(date, site, value) 

在这种情况下，每年开始于月和月结束。

我想要得到value每个year的mean和sd在3 sites。

我做了以下

df1 <- df %>% 
    dplyr::mutate(year = ifelse(date < "2002-05-01", "2001-2002", 
           ifelse(date < "2003-05-01", "2002-2003", 
            ifelse(date < "2004-05-01", "2003-2004", 
              ifelse(date < "2005-05-01", "2004-2005", 
                ifelse(date < "2006-05-01", "2005-2006", 
                  ifelse(date < "2007-05-01", "2006-2007", 
                   ifelse(date < "2008-05-01", "2007-2008", NA)))))))) %>% 
    dplyr::select(site, year, value) %>% 
    dplyr::group_by(site, year) %>% 
    dplyr::summarise_each(funs(
    mean(.), 
    sd(.) 
)) 

它给了我我想要的东西。但是，如果我有30-50年的数据，则需要花费时间。另外，如果每个新的data.frame都有不同的开始月份，我需要每次修改ifelse()以指定年份ID，以便能够按year进行分组并进行不同的计算。

如果开始月份是月份以外的任何月份，是否有任何直接的方式指定yearID？

Answer 1

什么

library(dplyr) 
df %>% 
    group_by(year=cut(date, seq(as.Date("2001-05-01"), as.Date("2008-05-01"), "1 year"), include.lowest = TRUE), site) %>% 
    summarise(sd = sd(value), mean = mean(value)) 
# Source: local data frame [21 x 4] 
# Groups: year [?] 
# 
#   year site  sd  mean 
#  (fctr) (fctr) (dbl) (dbl) 
# 1 2001-05-01 Site_1 51.82622 182.5890 
# 2 2001-05-01 Site_2 63.33385 241.1260 
# 3 2001-05-01 Site_3 30.04042 120.1233 
# 4 2002-05-01 Site_1 51.66325 182.6658 
# 5 2002-05-01 Site_2 62.87470 236.4192 
# 6 2002-05-01 Site_3 28.54769 122.2329 
# 7 2003-05-01 Site_1 50.97588 179.0874 
# 8 2003-05-01 Site_2 63.48810 227.1230 
# 9 2003-05-01 Site_3 30.87933 120.4918 
# 10 2004-05-01 Site_1 53.19898 176.5589 
# ..  ... ...  ...  ... 

Answer 2

使用包lubridate你可以先添加列year像这样：

library(lubridate) 
df$year <- ifelse(month(ymd(df$date)) < 5, 
        paste(year(ymd(df$date))-1, year(ymd(df$date)), sep="-"), 
        paste(year(ymd(df$date)), year(ymd(df$date))+1, sep="-")) 



df %>% dplyr::select(site, year, value) %>% 
    dplyr::group_by(site, year) %>% 
    dplyr::summarise_each(funs(
     mean(.), 
     sd(.) 
    )) 

Source: local data frame [6 x 4] 
Groups: site [1] 

    site  year  mean  sd 
    (fctr)  (chr) (dbl) (dbl) 
1 Site_1 2001-2002 178.2055 54.58277 
2 Site_1 2002-2003 176.9342 49.64435 
3 Site_1 2003-2004 177.4153 52.20447 
4 Site_1 2004-2005 179.5370 52.77848 
5 Site_1 2005-2006 180.3671 51.41292 
6 Site_1 2006-2007 179.3616 53.02291