equiwidth

You create equi-width (all bin have the same distance) histogram with the apps/search/search.jsp function.

WIDTH_BUCKET(column, min, max, bucket Number)

where:

• column is a date or number column
• From the min, the max and the bucket number the boundary of the buckets can be calculated.

Date Equi-width histogram Example where:

• the 'dataset' with clause represents your sample
• 7 is the number of day for your bucket
• 20 is the number of bucket

The below query gives you a histogram result that you can plot in Excel for instance.

WITH
  dataset AS
  (
    SELECT
      myColumnName AS Analyzed_Column -- Date or numeric
    FROM
      myTable
    WHERE
      myDateFilter > sysdate - 7*20 -- If needed
  )
  ,
  MINMAX AS
  (
    SELECT
      MIN(Analyzed_Column) AS min_value,
      MAX(Analyzed_Column) AS Max_Value
    FROM
      dataset
  )
  ,
  buckets AS
  (
    SELECT
      Analyzed_Column,
      WIDTH_BUCKET(Analyzed_Column, MINMAX.min_value, MINMAX.max_value, 20) AS
      bucket
    FROM
      dataset,
      MINMAX
  )
  ,
  buckets_full AS
  (
    SELECT
      rownum AS bucket
    FROM
      (
        SELECT
          1 AS bucket_number
        FROM
          dual
          CONNECT BY LEVEL <= (20 + 1)
      )
  )
  ,
  buckets_analysis AS
  (
    SELECT
      bucket,
      MAX(Analyzed_Column) AS max_value,
      MIN(Analyzed_Column) AS min_value,
      COUNT(1)         AS COUNT
    FROM
      buckets
    GROUP BY
      bucket
  )
SELECT
  buckets_full.bucket,
  buckets_analysis.max_value,
  buckets_analysis.min_value,
  buckets_analysis.count
FROM
  buckets_full
LEFT OUTER JOIN buckets_analysis
ON
  buckets_full.bucket = buckets_analysis.bucket
order by buckets_full.bucket;

equiheight

apps/search/search.jsp: equiheight histograms. Each bin get the same amount of records. The bin boundary varies.

See Quantile - Percentile

Example of output for a sample of 40 rows