获取两个日期之间的平均值

Question

我想计算两个日期之间变量的平均值。

这里是数据框的例子

library(lubridate) #ymd function

day= rep(seq.Date(from=ymd("2020-03-01"),to=ymd("2020-04-15"),by="day"), times=4)
center= rep(c("A", "B", "C", "D"), each=46)
ocupation= as.numeric(round(runif(184,20,40),1))
df <- data.frame(day,center,ocupation)


start <- mdy("03/15/2020","04/12/2020","05/01/2020","02/13/2020")
end <- mdy("03/20/2020","04/28/2020","05/14/2020","03/01/2020")
center<-c("A", "A", "B", "C")
id<-c(1,2,3,4)
patients <- data.frame(id, center,start,end)

显示的患者数据框只是一个样本，原始包含超过 12.000 个 ids

从每个 id 中，我想得到中心的开始日期和结束日期之间的平均职业

Answer 1

您可以使用来自tidyverse 的dplyr 包来执行此操作。

df <- as_tibble(df)

library(dplyr) # 1.0.0

df %>%
        # find only the days in df corresponding to day ranges in patients
        filter(day %in% c(seq(patients[1, 3], patients[1, 4], by = "days"),
                         seq(patients[2, 3], patients[2, 4], by = "days"),
                         seq(patients[3, 3], patients[3, 4], by = "days"),
                         seq(patients[4, 3], patients[4, 4], by = "days"))) %>%
        # add id column
        mutate(id = ifelse(day %in% seq(patients[1, 3], patients[1, 4], by = "days"), patients$id[1],
                           ifelse(day %in% seq(patients[2, 3], patients[2, 4], by = "days"), patients$id[2],
                                  ifelse(day %in% seq(patients[3, 3], patients[3, 4], by = "days"), patients$id[3], patients$id[4])))) %>%
        # group by id
        group_by(id) %>%
        # find mean occuption for each id
        summarise(mean_occupation = mean(ocupation))

# A tibble: 3 x 2
     id mean_occupation
  <dbl>           <dbl>
1     1            29.7
2     2            31.7
3     4            32.2

编辑

带有for 的版本会为许多id 循环：

df <- as_tibble(df)
library(dplyr)

# create days vector from patients
days <- list()
for (i in 1:nrow(patients)) {
        dates <- seq(patients[i, 3], patients[i, 4], by = "days")
        for (j in 1:length(dates)) {
                names(dates)[j] <- patients$id[i]
        }
        days[[i]] <- dates
}
days <- as.Date(unlist(days), origin = "1970-01-01")

# filter df for days
mid <- df %>%
        filter(day %in% days)

# create id col (I couldn't do this directly in mutate())
id <- character()
for (i in 1:nrow(mid)) {
        id[i] <- names(days)[which(days == mid$day[i])]
}

# bind together and finish
final <- mid %>%
        cbind(id) %>% as_tibble() %>%
        group_by(id) %>%
        summarise(mean_occupation = mean(ocupation))

> final
# A tibble: 3 x 2
  id    mean_occupation
  <chr>           <dbl>
1 1                29.7
2 2                31.7
3 4                32.2

Answer 2

我会创建一个函数来返回一个 id 的平均占用率：

mean.occ = function(id, patients, occupency, day, center){
  to.select = day > patients[id, "start"] & day < patients[id, "end"] & center == patients[id, "center"]
  return(mean(occupency[to.select]))
}

这里，day > patients[id, "start"] & day < patients[id, "end"] & center == patients[id, "center"] 选择特定 id 的开始日期和结束日期之间的占用率值，并对应于给定的中心。

然后使用sapply 将其应用于每个id：

mean.occupancies = sapply(patients$id, FUN = mean.occ, patients, ocupation, day, center)

终于可以将结果添加到patients数据框了：

patients = cbind.data.frame(patients, mean.occupancies)