Question

我想用滚动窗口计算11个月的总和，使用每日数据。数据框是这样的。我有17年的每日数据，该文件中包含300多家公司

head(Rolling_window)
# A tibble: 6 x 5
        date price1 price2 price3 price4
      <dttm>  <dbl>  <dbl>  <dbl>  <dbl>
1 2000-01-04  72.27   6.63   2.38   2.98
2 2000-01-06  72.27   6.63   2.39   2.98
3 2000-01-11  72.27   6.63   2.41   2.98
4 2000-01-12  69.00   6.63   2.41   2.98
5 2000-01-13  68.67   6.63   2.41   2.98
6 2000-01-14  69.60   6.63   2.41   2.98

Answer 1

如果我理解你，你可以使用这个解决方案：

library(tidyverse)

date <- c('2000-01-04','2000-01-06','2000-01-11','2000-01-12',
          '2000-01-13','2000-01-14','2000-02-14','2001-02-14')
n <- NROW(date)
df <- tibble(date,price1 = rnorm(n),price2 = rnorm(n,0.1),price3 = rnorm(n,10),price4 = rnorm(n,5))
df %>% 
  mutate(date = as.character(date)) %>% 
  separate('date',c('year','month','day')) %>% 
  group_by(year,month) %>% 
  summarise_at(vars(price1, price2, price3, price4),funs(mean)) %>% 
  ungroup

快速更新

我使用data.table包重写代码，现在它的工作速度会快得多：

require(data.table)

df <- as.data.table(df)
df_sum <- df[, c('year','month','day') := tstrsplit(date, "-", fixed=TRUE)
             ][
               ,c('date','day'):=NULL,
               ][
                 , lapply(.SD, mean), by = list(year,month)
                 ][
                   , rollapply(.SD,11,sum,partial = T),
                     .SDcols=c('price1','price2','price3','price4')
                    ]

Answer 2

请在下面找到以下代码

library(dplyr)
library(zoo)
library(lubridate)
dummy = data.frame(Date = seq(as.Date("2016-01-01"),as.Date("2017-12-31"),"day"),
                   price1 = sample(5.5:9.5,731,replace = T),
                   price2 = sample(5.5:9.5,731,replace = T),
                   price3 = sample(5.5:9.5,731,replace = T),stringsAsFactors = F)

dummy_rolling_11m = dummy %>% mutate(yearmonth = as.yearmon(Date)) %>% 
  select(-Date)%>% group_by(yearmonth) %>%
  summarise_all(funs(sum(.,na.rm = T))) %>% 
  arrange(yearmonth) %>% select(-yearmonth) %>% 
  mutate_all(funs(rollapplyr(.,11,sum,partial = T)))

每月滚动窗口总和来自r

2 个答案: