在 SQL 中转置/旋转数据

Question

I'm sorry to ask yet another question that pertains to this topic—I read through the other threads but I still can not figure out my specific case.

So I have an output:

The query that outputs the above is:

SELECT  (CASE
            WHEN ca.iteration = 'Pre' THEN '1. Pre'
            WHEN ca.iteration = 'Post-2' THEN '3. Post-2'
            WHEN ca.iteration = 'Post' THEN '2. Post'
            END) AS capes_iteration,
           ROUND(AVG(ca.emotional_problems_score)) AS avg_child_emotions_score,
           ROUND(AVG(ca.behavioral_problems_score)) AS avg_child_behaviors_score,
           ROUND(AVG(ca.total_score)) AS avg_total_intensity_score,
           ROUND(AVG(ca.total_parent)) AS avg_parent_confidence_score
     FROM  reporting_f5scppp.capes ca
LEFT JOIN  public.contacts c ON c.id = ca.contact_id
 GROUP BY  capes_iteration;

What I want it to look like is this:

I've spent about 4 hours trying to get this to no avail—I'm only a beginner to SQL—seeing a solution to this will be a great learning opportunity for me.

If anyone can provide a simple solution—lunch is on me!

Answer 1

You can unpivot the scores using a lateral join. Then just reaggregate:

SELECT v.which,
       AVG(v.score) FILTER (WHERE ca.iteration = 'Pre') as avg_1_pre,
       AVG(v.score) FILTER (WHERE ca.iteration = 'Post-2') as avg_3_post_2,
       AVG(v.score) FILTER (WHERE ca.iteration = 'Post') as avg_2_post
FROM reporting_f5scppp.capes ca LEFT JOIN
     public.contacts c
     ON c.id = ca.contact_id LEFT JOIN LATERAL
     (VALUES (emotional_problems_score, 'emotional'),
             (behavioral_problems_score, 'behaviors'),
             (total_score, 'intensity'),
             (total_parent, 'parent')
     ) v(score, which)
GROUP BY  v.which;

Answer 2

One option would be unpivoting by using each() function in order to get key-value tuples, and then pivoting by conditional aggregation in the main query:

SELECT (kv).key AS capes_iteration, 
       MAX(CASE WHEN iteration = 'Pre' THEN (kv).value END) AS "1. Pre",
       MAX(CASE WHEN iteration = 'Post-2' THEN (kv).value END) AS "3. Post-2",
       MAX(CASE WHEN iteration = 'Post' THEN (kv).value END) AS "2. Post"
  FROM
  (
   SELECT iteration, each(hstore(q) - 'iteration'::text) AS kv
     FROM (
           SELECT iteration,
                  ROUND(AVG(COALESCE(ca.emotional_problems_score,0))) AS avg_child_emotions_score,
                  ROUND(AVG(COALESCE(ca.behavioral_problems_score,0))) AS avg_child_behaviors_score,
                  ROUND(AVG(COALESCE(ca.emotional_problems_score,0)+
                            COALESCE(behavioral_problems_score,0))) AS avg_total_intensity_score,
                  ROUND(AVG(ca.total_parent)) AS avg_tot_parent_confidence_score
             FROM capes ca
             LEFT JOIN  contacts c ON c.id = ca.contact_id
            GROUP BY  iteration ) q ) q2
 GROUP BY capes_iteration
 ORDER BY capes_iteration

where

• use COALESCE() function with zero as the second argument to assign the null values to zero
• notice that avg_total_intensity_score is a derived column from addition of two columns

Demo