Question

I need to get maximum and minimum values but also I need to get row id of these maximum or minimum on the same row.

SELECT MIN([Value]), MAX([Value]), id 
FROM [AnalystEstimates].[dbo].[AnalystEstimateValues] 
GROUP BY indicatorid

Answer 1

It’s very unclear what you want from your question. Do you really want the GROUP BY indicatorid? If not then it’s quite simple and you already have many answers. But if you do want to GROUP BY then it’s more difficult and no-one has got it quite right yet. I also assume that you only want one row per indicatorid, and if there are duplicate rows that have the same max/min then it’s better to just choose one of them arbitrarily instead of returning both.

Here’s my attempt, using CTEs (requires SQL Server 2005 or newer):

WITH
    RowNumbers AS (
        SELECT ROW_NUMBER() OVER (ORDER BY indicatorid, value) AS RowNumber, *
        FROM [AnalystEstimates].[dbo].[AnalystEstimateValues]),
    MinRowNumbers AS (
        SELECT indicatorid, MIN(RowNumber) AS RowNumber FROM RowNumbers GROUP BY indicatorid),
    MaxRowNumbers AS (
        SELECT indicatorid, MAX(RowNumber) AS RowNumber FROM RowNumbers GROUP BY indicatorid)
SELECT
    MinRowNumbers.indicatorid,
    RN1.Value AS MinValue,
    RN1.ID AS MinValueId,
    RN2.Value AS MaxValue,
    RN2.ID AS MaxValueId
FROM MinRowNumbers
JOIN MaxRowNumbers ON MinRowNumbers.indicatorid = MaxRowNumbers.indicatorid
JOIN RowNumbers RN1 ON MinRowNumbers.RowNumber = RN1.RowNumber
JOIN RowNumbers RN2 ON MaxRowNumbers.RowNumber = RN2.RowNumber

Here is some data I used to test it:

CREATE TABLE AnalystEstimateValues (ID int, indicatorid int, Value int);

INSERT INTO AnalystEstimateValues (ID, indicatorid , Value) VALUES
(1, 1, 4),
(2, 1, 4),
(3, 2, 6),
(4, 1, 2),
(5, 2, 2),
(6, 2, 5),
(7, 3, 0);

And here’s the output I get:

indicatorid MinValue MinValueId MaxValue MaxValueId
          1        2          4        4          2
          2        2          5        6          3
          3        0          7        0          7

If this isn’t what you want, can you please try to improve your question to tell us what you do want?

---

Update: Here’s an alternative solution based on Craig Young’s answer but using joins instead of subselects:

WITH
    UniqueIds AS (
        SELECT IndicatorId, Value, MIN(id) AS Id
        FROM AnalystEstimateValues
        GROUP BY IndicatorId, Value)
SELECT
    lims.IndicatorId,
    MinValue,
    T1.Id AS MinValueId,
    MaxValue,
    T2.Id AS MaxValueId 
FROM (
        SELECT
            IndicatorId,
            MIN(Value) as MinValue,
            MAX(Value) as MaxValue
        FROM AnalystEstimateValues
        GROUP BY IndicatorId) lims
JOIN UniqueIds T1 ON lims.IndicatorId = T1.IndicatorId AND lims.MinValue = T1.Value
JOIN UniqueIds T2 ON lims.IndicatorId = T2.IndicatorId AND lims.MaxValue = T2.Value

This is cleaner and probably also faster than my first version, although I haven’t run performance tests to verify this.