从 R 中的年、月和周数据插入年-月-日

Question

I've inherited a dataset with measurements spanning 1970–2019. The head and tail look something like this:

year  month  week    X1 
1970      1     1   0.21
1970      1     2   0.22
1970      1     3   0.34
1970      1     4   0.34
1970      2     5   0.35
1970      2     6   0.25
... 
2019     11    47   0.063
2019     12    48   0.062
2019     12    49   0.068
2019     12    50   0.067
2019     12    51   0.074
2019     12    52   0.075

Each observation of X1 was recorded on the first day of each week (ie, Monday). I'd like to create a date column in ISO 8601 format (yyyy-mm-dd). Given year, month, and week, it should be possible to extract which day of the month the Monday of each week is. Note: measurements were taken every Monday, regardless of holidays.

Answer 1

You can use base R:

df <- data.frame(
  year = c(1970,1970,1970,1970,1970,1970,2019,2019,2019,2019),
  month = c(1,1,1,1,2,2,11,12,12,12),
  week = c(1,2,3,4,5,6,47,48,49,50)
)

df$date_string <- paste(df$year,df$week,1, sep = "-")
df$date <- as.Date(x = df$date_string,format = "%Y-%U-%u")

You can have a look at: https://www.rdocumentation.org/packages/base/versions/3.6.2/topics/strptime

'%U' converts the week of the year and the '1' is needed for the first day of the week.

Answer 2

This is really just a one-liner. You can generate a vector of every Monday since 5th January 1970 using the lubridate package like this:

as.POSIXct("1970-01-05") + lubridate::days(0:2616 * 7)

That takes you up to today.

Here's a reprex showing the first 100 Mondays since the start of 1970:

head(as.POSIXct("1970-01-05") + lubridate::days(0:2616 * 7), 100)
#>   [1] "1970-01-05 BST" "1970-01-12 BST" "1970-01-19 BST" "1970-01-26 BST"
#>   [5] "1970-02-02 BST" "1970-02-09 BST" "1970-02-16 BST" "1970-02-23 BST"
#>   [9] "1970-03-02 BST" "1970-03-09 BST" "1970-03-16 BST" "1970-03-23 BST"
#>  [13] "1970-03-30 BST" "1970-04-06 BST" "1970-04-13 BST" "1970-04-20 BST"
#>  [17] "1970-04-27 BST" "1970-05-04 BST" "1970-05-11 BST" "1970-05-18 BST"
#>  [21] "1970-05-25 BST" "1970-06-01 BST" "1970-06-08 BST" "1970-06-15 BST"
#>  [25] "1970-06-22 BST" "1970-06-29 BST" "1970-07-06 BST" "1970-07-13 BST"
#>  [29] "1970-07-20 BST" "1970-07-27 BST" "1970-08-03 BST" "1970-08-10 BST"
#>  [33] "1970-08-17 BST" "1970-08-24 BST" "1970-08-31 BST" "1970-09-07 BST"
#>  [37] "1970-09-14 BST" "1970-09-21 BST" "1970-09-28 BST" "1970-10-05 BST"
#>  [41] "1970-10-12 BST" "1970-10-19 BST" "1970-10-26 BST" "1970-11-02 BST"
#>  [45] "1970-11-09 BST" "1970-11-16 BST" "1970-11-23 BST" "1970-11-30 BST"
#>  [49] "1970-12-07 BST" "1970-12-14 BST" "1970-12-21 BST" "1970-12-28 BST"
#>  [53] "1971-01-04 BST" "1971-01-11 BST" "1971-01-18 BST" "1971-01-25 BST"
#>  [57] "1971-02-01 BST" "1971-02-08 BST" "1971-02-15 BST" "1971-02-22 BST"
#>  [61] "1971-03-01 BST" "1971-03-08 BST" "1971-03-15 BST" "1971-03-22 BST"
#>  [65] "1971-03-29 BST" "1971-04-05 BST" "1971-04-12 BST" "1971-04-19 BST"
#>  [69] "1971-04-26 BST" "1971-05-03 BST" "1971-05-10 BST" "1971-05-17 BST"
#>  [73] "1971-05-24 BST" "1971-05-31 BST" "1971-06-07 BST" "1971-06-14 BST"
#>  [77] "1971-06-21 BST" "1971-06-28 BST" "1971-07-05 BST" "1971-07-12 BST"
#>  [81] "1971-07-19 BST" "1971-07-26 BST" "1971-08-02 BST" "1971-08-09 BST"
#>  [85] "1971-08-16 BST" "1971-08-23 BST" "1971-08-30 BST" "1971-09-06 BST"
#>  [89] "1971-09-13 BST" "1971-09-20 BST" "1971-09-27 BST" "1971-10-04 BST"
#>  [93] "1971-10-11 BST" "1971-10-18 BST" "1971-10-25 BST" "1971-11-01 GMT"
#>  [97] "1971-11-08 GMT" "1971-11-15 GMT" "1971-11-22 GMT" "1971-11-29 GMT"

^{Created on 2020-02-24 by the reprex package (v0.3.0)}

Answer 3

using lubridate package, you can calculate as following:

df <- data.frame(
  year = c(1970,1970,1970,1970,1970,1970,2019,2019,2019,2019),
  month = c(1,1,1,1,2,2,11,12,12,12),
  week = c(1,2,3,4,5,6,47,48,49,50)
)

df$year_first_day <- lubridate::ymd(paste(df$year, '0101', sep = ''))
df$year_first_monday <- lubridate::ceiling_date(df$year_first_day, unit = 'weeks', week_start = 1)
df$date <- lubridate::dweeks(df$week - 1) + df$year_first_monday
df
#    year month week year_first_monday year_first_day       date
# 1  1970     1    1        1970-01-05     1970-01-01 1970-01-05
# 2  1970     1    2        1970-01-05     1970-01-01 1970-01-12
# 3  1970     1    3        1970-01-05     1970-01-01 1970-01-19
# 4  1970     1    4        1970-01-05     1970-01-01 1970-01-26
# 5  1970     2    5        1970-01-05     1970-01-01 1970-02-02
# 6  1970     2    6        1970-01-05     1970-01-01 1970-02-09
# 7  2019    11   47        2019-01-07     2019-01-01 2019-11-25
# 8  2019    12   48        2019-01-07     2019-01-01 2019-12-02
# 9  2019    12   49        2019-01-07     2019-01-01 2019-12-09
# 10 2019    12   50        2019-01-07     2019-01-01 2019-12-16

Answer 4

Here is one idea. Notice that I only used the first six rows from your example for this demonstration.

library(dplyr)
library(lubridate)

date_seq <- tibble(
  # Create a data frame with dates from 1970 to 2019
  date = seq.Date(as.Date("1970-01-01"), as.Date("2019-12-31"), by = 1)
) %>%
  # Create weekday
  mutate(weekday = weekdays(date)) %>%
  # Filter for Monday
  filter(weekday %in% "Monday") %>%
  # Create year, month
  mutate(year = year(date), month = month(date)) %>%
  # Create week number
  mutate(week = 1:n()) %>%
  # Join the data
  left_join(dat, by = c("year", "month", "week"))
date_seq
# # A tibble: 2,609 x 6
#    date       weekday  year month  week    X1
#    <date>     <chr>   <dbl> <dbl> <int> <dbl>
#  1 1970-01-05 Monday   1970     1     1  0.21
#  2 1970-01-12 Monday   1970     1     2  0.22
#  3 1970-01-19 Monday   1970     1     3  0.34
#  4 1970-01-26 Monday   1970     1     4  0.34
#  5 1970-02-02 Monday   1970     2     5  0.35
#  6 1970-02-09 Monday   1970     2     6  0.25
#  7 1970-02-16 Monday   1970     2     7 NA   
#  8 1970-02-23 Monday   1970     2     8 NA   
#  9 1970-03-02 Monday   1970     3     9 NA   
# 10 1970-03-09 Monday   1970     3    10 NA   
# # ... with 2,599 more rows

DATA

dat <- read.table(text = "year  month  week    X1 
1970      1     1   0.21
1970      1     2   0.22
1970      1     3   0.34
1970      1     4   0.34
1970      2     5   0.35
1970      2     6   0.25",
                header = TRUE, stringsAsFactors = FALSE)