I’ve written R code to produce a periodic report that requires re-ordering of Week numbers such that I can filter and order by the most recent 10 weeks. To prevent errors and minimize hard-coded values, I prefer to declare this week order at the top of the script that sources the other several scripts used. Thus, I would like to define an ordered factor list and then use it to order the week number column later. RepEx below, but generally I am reordering all 52 weeks such that the most recent 10-week-period is last/largest, e.g. new_levels <- factor(1:52, levels = c(29:52, 1:28), ordered=TRUE).
Side note: any advise on how better to handle grabbing the most recent (not necessarily greatest) 10-week period is welcomed. My struggle in the past is due to the roll-over near the end of the year (51, 52, 1, 2, 3,…).
Example:
new_levels <- factor(1:10, levels = c(8:10, 1:7), ordered=TRUE)
data <- tibble(Week = 1:10, ID = c("A","A","B","B","C","A","D","B","D","A"))
data <- data %>% mutate(Week2 = factor(Week, levels = new_levels, ordered = TRUE)) %>% arrange(Week2)
The ordered factor (new_levels) appears to be correct, but the behavior of arrange() and str() show that the ordering I want is not happening:
> new_levels
[1] 1 2 3 4 5 6 7 8 9 10
Levels: 8 < 9 < 10 < 1 < 2 < 3 < 4 < 5 < 6 < 7
> data
# A tibble: 10 × 3
Week ID Week2
<int> <chr> <ord>
1 1 A 1
2 2 A 2
3 3 B 3
4 4 B 4
5 5 C 5
6 6 A 6
7 7 D 7
8 8 B 8
9 9 D 9
10 10 A 10
> str(data)
tibble [10 × 3] (S3: tbl_df/tbl/data.frame)
$ Week : int [1:10] 1 2 3 4 5 6 7 8 9 10
$ ID : chr [1:10] "A" "A" "B" "B" ...
$ Week2: Ord.factor w/ 10 levels "1"<"2"<"3"<"4"<..: 1 2 3 4 5 6 7 8 9 10
Thank you!
>Solution :
If you look closer at your output, you will see you are not doing what you are expecting:
data %>%
mutate(Week2 = factor(Week, levels = new_levels, ordered = TRUE)) %>%
pull(Week2)
# [1] 1 2 3 4 5 6 7 8 9 10
# Levels: 1 < 2 < 3 < 4 < 5 < 6 < 7 < 8 < 9 < 10
This shows that arrange is working as expected. The issue comes from the fact that you are assigning levels = new_levels. What is the value of new_levels?
new_levels
# [1] 1 2 3 4 5 6 7 8 9 10
# Levels: 8 < 9 < 10 < 1 < 2 < 3 < 4 < 5 < 6 < 7
In this case it is a sequence of 1:10. What you want is to assign the levels of new_levels to the levels of your new variable:
data %>%
mutate(Week2 = factor(Week, levels = levels(new_levels), ordered = TRUE)) %>%
arrange(Week2)
# Week ID Week2
# <int> <chr> <ord>
# 1 8 B 8
# 2 9 D 9
# 3 10 A 10
# 4 1 A 1
# 5 2 A 2
# 6 3 B 3
# 7 4 B 4
# 8 5 C 5
# 9 6 A 6
# 10 7 D 7