如何通过R中的组/因子计算NA值的长度？

Question

I am tasked with manipulating data obtained from 1258 unique surveys.

In terms of dimensions. 28 million individual observations (including NA) -8 columns (variables). object name : dat

The column/variable I am particularly interested in is education (edu). I want to get the length of NA and Non-NA values (for edu) for those studies by aggregating (data$ edu ~ id_study )

So far, I have used this code to work out the number of studies which contain at least 1 or more entries on edu.

numbers <- aggregate(dat$edu ~ dat$id_study, data=dat, FUN=length)

I have the result I need for quantifying the numbers of unique id_study that have data on edu. This ticks box one.

Now I need to do the same for the unique id_study that have nothing at all on education. How do i do this?

I've tried so many codes to work out the length of NAs for studies that do not have anything on edu.

aggregate_2 <- aggregate(dat$edu ~ id_study, data=dat, FUN=length(dat[!is.na(dat)]))

this does not work :(

Can anyone shed some light on this matter please?

thank you

EDIT ****** Just to clarify if i was not clear in my question. There are 1258 unique surveys/studies,(some surveys may be for multiple years, eg ALB_2013 and ALB_2014 under id_study).

Out of these surveys, using equation 1 code and the code i put in the description, code 1 , I worked out that 530 of these 1258 surveys provided >=1 individual observation under the edu column.

This must mean 728 Unique surveys did not provide any information at all under the edu. I want to work out the names of the 728 surveys and using a function, hopefully want to work out the length of NAs per survey which didn't provide any information at all.

I hope this makes sense.

id_study (name of the survey) id (survey id) column i'm interested in is "edu".

Answer 1

First off: Posting a screenshot of your data is bad practice, as it would require SO respondents to manually type in your sample data. Use dput to post (part of) your data. For future questions, follow the advice and links in Sotos' first comment!

That aside, how about the following:

numbers <- aggregate(
    edu ~ id_study, 
    data = dat, 
    FUN = function(x) c(n_nonNA = sum(!is.na(x)), n_NA = sum(is.na(x))))