RSQLite using group-by for functions within the query-CodePudding

I am very new to SQL (using it in R currently with RSQLite and DBI packages)

I am trying to create a column that is the absolute mean deviation, aka:

(i) - AVG(i,g)

Where i is the individual occurence and the AVG component is the average for the group. What I am having troubles with is making sure the AVG component is the only part that gets grouped. When I do GROUP BY, it groups everything and doesn't give me the right number.

Here is the sample data:

student	class	grade
A	English	79
A	Spanish	65
A	Chemistry	92
B	English	46
B	Spanish	83
B	Chemistry	78
C	English	67
C	Spanish	87
C	Chemistry	98
D	English	99
D	Spanish	80
D	Chemistry	75

Basically I would want the individual GRADE for a student in a class to compare with the average of that student (e.g. the individual english grade - the total average for a student)

Iha ve tried the following:

dbGetQuery(gradesdb, "SELECT student, 
                      ABS(grade-AVG(grade)) AS mad
                      FROM grades
                      GROUP BY student,class")

This gives me 0 for all of the stat values (which I gather is because the group by is going on all selected operations within the query)

how can I make it so that the AVG portion of the calculation is only "grouped" by the student. I get the right calculation if I do:

dbGetQuery(gradesdb2, "SELECT student, 
                      ABS(grade-AVG(grade)) AS mad
                      FROM grades
                      GROUP BY student")

But then I only get the first class for each student, instead of the stat for each class with the student average.

I want to do this all in SQL and not calculate the average as a seperate column with base R or tidyverse.

Thank you so much for your help!

CodePudding user response：

Consider turning AVG() via GROUPBY to AVG() via a window function:

SELECT student, 
       ABS(grade - AVG(grade) OVER (PARTITION BY student)) AS mad
FROM grades