Cf add industry by subject dropdown

[chfoster]

Pull request overview

Add extra dropdown to the industry by subject tables, close #42.

Pull request checklist

Please check if your PR fulfils the following:

• [ ] Tests for the changes have been added (for bug fixes / features)
• [ ] Docs have been reviewed and added / updated if needed (for bug fixes / features)
• [x] Tests have been run locally and are passing (run_tests_locally())
• [x] Code is styled according to tidyverse styling (checked locally with tidy_code())

What is the new behaviour?

A new dropdown has been added to the industry by subject tables that allows users to filter the tables further - to the 3-digit SIC group level rather than just the SECTIONNAME level. The dropdown updates based on the selected SECTIONNAME to only include the appropriate group_names.

Anything else

@rmbielby to look at allowing multiple selections at a later date.

[rmbielby]

Quick suggestion to speed the code up a little: Whenever you have a group_by()/summarise() combo, it's worth adding .groups="drop" to the summarise as it generally halves the time it takes to run the summarise() function (if you don't set it, summarise just defaults to it anyway, but takes a bit of extra time to decide that). As well, you're never doing any actual sums in the summarises, so the lines where you've got n=sum(count), you should be able to change to n=count.

So for example, the following: group_by(sex, subject_name) %>% summarise(n = sum(count)) %>%

Should run quicker, but produce the same result, if changed to:

group_by(sex, subject_name) %>% summarise(n = count, .groups = "drop") %>%

[chfoster]

Quick suggestion to speed the code up a little: Whenever you have a group_by()/summarise() combo, it's worth adding .groups="drop" to the summarise as it generally halves the time it takes to run the summarise() function (if you don't set it, summarise just defaults to it anyway, but takes a bit of extra time to decide that). As well, you're never doing any actual sums in the summarises, so the lines where you've got n=sum(count), you should be able to change to n=count.

So for example, the following: group_by(sex, subject_name) %>% summarise(n = sum(count)) %>%

Should run quicker, but produce the same result, if changed to:

group_by(sex, subject_name) %>% summarise(n = count, .groups = "drop") %>%

Do you know any lines specifically where this could be changed? I can see it in the regional analysis but I do need to sum there. In the industry_subject and subject_industry I think it is already changed?

[rmbielby]

Quick suggestion to speed the code up a little: Whenever you have a group_by()/summarise() combo, it's worth adding .groups="drop" to the summarise as it generally halves the time it takes to run the summarise() function (if you don't set it, summarise just defaults to it anyway, but takes a bit of extra time to decide that). As well, you're never doing any actual sums in the summarises, so the lines where you've got n=sum(count), you should be able to change to n=count. So for example, the following: group_by(sex, subject_name) %>% summarise(n = sum(count)) %>% Should run quicker, but produce the same result, if changed to: group_by(sex, subject_name) %>% summarise(n = count, .groups = "drop") %>%

Do you know any lines specifically where this could be changed? I can see it in the regional analysis but I do need to sum there. In the industry_subject and subject_industry I think it is already changed?

Yep, was just meaning in the industry by subject code. Lines 368, 388, 421, 447 and so on all either have n=sum(count) when there's no summation going on (as far as I could tell, but correct me if I'm wrong!), or have summarise(n=earnings_median) without a .groups="drop" flag.