Frequency of column name lengths based on the table type

Copy SQL

WITH columns AS (SELECT nspname AS table_schema, relname AS table_name, 
CASE WHEN relkind='r' THEN 'BASE TABLE' 
WHEN relkind='m' THEN 'MATERIALIZED VIEW' 
ELSE 'VIEW' END AS table_type, attname AS column_name
FROM pg_class INNER JOIN pg_namespace ON pg_class.relnamespace=pg_namespace.oid
INNER JOIN pg_attribute ON pg_class.oid=pg_attribute.attrelid
WHERE relkind IN ('r', 'm','v') AND attnum>=1 AND 
nspname NOT IN (SELECT schema_name
FROM INFORMATION_SCHEMA.schemata
WHERE schema_name<>'public' AND
schema_owner='postgres' AND schema_name IS NOT NULL))
SELECT table_type, length(column_name) AS length, Count(*) AS nr_of_occurrences, Round(Count(*)::decimal*100/(SELECT Count(*) FROM columns AS c2 WHERE c2.table_type=columns.table_type),1) AS percentage_of_all_columns
FROM columns
GROUP BY table_type, length(column_name)
ORDER BY table_type, length;