Business Intelligence Fundamentals

Objectives

In this session, you will learn to:

Create MDX queries
Use MDX functions for analysis

Slide 1 of 27

Ver 1.0

Business Intelligence Fundamentals

Difference Between SQL and MDX

The MDX query language is similar to Structured Query

Language (SQL).
It enables you to query multidimensional objects, such as
OLAP cubes.
The following table shows some of the common differences
between SQL and MDX.
SQL Queries

MDX Queries

Work with two dimensions

Use the SELECT statement to define
the column layout and the WHERE
clause to define the row layout

Work with multiple dimensions

Use the SELECT clause to define
multiple axes and the WHERE clause to
restrict the data to a specific dimension
or measure
Use the WHERE clause to identify a
portion of data from a different dimension

Use the WHERE clause to filter data

Return a subset of two-dimensional
data from tables
Populate a predefined
two-dimensional rowset if they are
created on one or more tables
Access the SQL Server database that
can be connected from client
applications

Return a subset of multidimensional data

from cubes
Populate a multidimensional dataset if
they are created on a single cube
Access the OLAP cube data that can be
connected with the cube browser using
ADOMD and OLE DB for OLAP

Slide 2 of 27

Ver 1.0

Business Intelligence Fundamentals

Difference Between SQL and MDX (Contd.)

The following figure shows the difference in connectivity

between the SQL Server database and the OLAP cube.

Slide 3 of 27

Ver 1.0

Business Intelligence Fundamentals

Key Concept in MDX

An OLAP cube consists of dimensions and measures.

In the figure below, there are two dimensions: Time and
Geography. The Geography dimension has five members: Africa,
Asia, Australia/Pacific, Europe, and North America. There is
one measure, QuantitySold, that is the number of items sold in
each continent by year.

QuantitySold
Time
Geography
Africa
Asia
Australia/Pacific
Europe
North America
4

1998

1999

2000

2001

2002

10

1429

1188

1242

1162

2133

23654

27184

31359

26476

31069

6726

8484

11032

8616

8343

Slide 4 of 27

Ver 1.0

Business Intelligence Fundamentals

Key Concepts in MDX (Contd.)

A cell is the summarized values contained in the intersection of two

or more levels.
In a two-dimensional view of a cube, each cell can be described in
terms of one member from each dimension. For example, the
number of items sold in North America in 2002 was 8,343.

QuantitySold
Time
Geography
Africa
Asia
Australia/Pacific
Europe
5

North America

1998

1999

2000

2001

2002

10

1429

1188

1242

1162

2133

23654 27184 31359 26476 31069

6726

8484 11032
Slide 5 of 27

8616

8343

Ver 1.0

Business Intelligence Fundamentals

Key Concepts in MDX (Contd.)

Dimensions in a cube are hierarchical. Values are derived by adding

or aggregating the original data in the cube to produce a
presummarized report.

QuantitySold
Africa Asia Australia Europe North
/
America
Pacific
2002 Q1

358

4772

1261

Q2

392

8564

2307

Q3

380

9811

2653

Oct

182

1395

412

Nov

257

1865

556

Dec

564

4663

1154

Q4 Total

1003

7922

2122

Total

27
2 Slide 6 of 2133

31069

8343

Q4

2002

Ver 1.0

Business Intelligence Fundamentals

Tuples

To identify and extract data, MDX uses a reference system named

tuples. Tuples list dimensions and members to identify individual
cells, as well as a larger section of cells in the cube. Because each
cell is an intersection of all the dimensions of the cube, tuples
uniquely
identify
every
Product
(Jun 2001, United States, Clothes)
cell in
Dimension
the cube.
Geography
Dimension

Clothes

United
States
Time Dimension
7

Jun 2001

Slide 7 of 27

Ver 1.0

Business Intelligence Fundamentals

Tuples (Contd.)

Tuples that encompass more than one cell are named slices.
1st Dimension:
Geography
Continent
Country
Region

Geography
Geography==Europe
Europe
Product
Product==All
AllProducts
Products
Time
Time==All
AllTime
Time

Cell
Cell
Cell
Cell
Cell
Cell
Cell
Cell
Cell
Cell
Cell
Cell
Cell
Cell
Cell

Slide 8 of 27

2nd Dimension:
Product
Line
Category
Group

3rd Dimension:
Time
Year
Quarter
Month
Ver 1.0

Business Intelligence Fundamentals

Naming Conventions

When writing MDX code, you need a way to identify the specific
members with which you want the code to work. You can identify a
member by working downwards from the dimension name and
specifying the members at each level until you reach the required
member. For example,
[Geography].[All Geo].[Africa].[Benin]
is used to indicate the Benin member.

All

Continent

All Geo

North
America

Africa

Country Benin

Morocco
Slide 9 of 27

Canada

United
States

Ver 1.0

Business Intelligence Fundamentals

MDX Syntax

[ ] The square brackets encircle cube names,

dimension names, hierarchy names, and member
names.
For example:
[Geography].[All Geo].[Africa].
[Benin]
.

10

The dot separates the member names and the

dimension names.

Slide 10 of 27

Ver 1.0

Business Intelligence Fundamentals

Sets

A set is a collection of tuples that were defined using the same

dimension. For example:

([Geography].[Benin], [Time].[2001].[Q1], [Product].[Clothes])

([Geography].[Canada], [Time].[2001].[Q1], [Product].[Clothes])

11

Both have exactly one member from the Geography, Time, and Product
dimensions. Therefore, these two tuples form a set when taken together.

Slide 11 of 27

Ver 1.0

Business Intelligence Fundamentals

MDX Syntax

( ) Parentheses encircle tuples. For example:

([Geography].[All Geo].[Africa].[Benin])

and
([Geography].[All Geo].[North America].[Canada])

{ } Curly brackets encircle sets. For example:

{([Geography].[All Geo].[Africa].[Benin]),
([Geography].[All Geo].[North America].[Canada])
}

12

The comma separates members within tuples and separates

tuples within sets.

The colon defines a range. For example:

{ [Time].[2001].[Q1]:[Time].[2001].[Q4] }

Slide 12 of 27

Ver 1.0

Business Intelligence Fundamentals

Axis and Slicer Dimensions

When you formulate an MDX query, an

Axis
determine the
edges at
of athe
multidimensional
application typically
looks
cubes and
dimensions
result set. Data
divides the set
of dimensions
into two subsets:
is retrieved
for multiple members.
Slicer
dimensions

13

filter multidimensional data. Data

is retrieved for a single member.

Slide 13 of 27

Ver 1.0

Business Intelligence Fundamentals

Slicers

14

Implicit
Slicers

are dimensions not explicitly assigned

to an axis. It is assumed that you want the
default hierarchy's default member for any
dimension that remains as an implicit slicer.

Explicit
Slicers

are derived by using the WHERE clause to

restrict multidimensional data, which limits the
data returned by a query.

Slide 14 of 27

Ver 1.0

Business Intelligence Fundamentals

Basic MDX Query Structure

The components that an MDX query needs in order

to extract the requested information are as follows:
a component to specify the column headers
a component to specify the row headers
a pointer to the cube being accessed
General form of the SELECT statement:

SELECT
SELECT
{set
{setdefining
definingthe
thecolumn
columnheaders}
headers} ON
ONCOLUMNS,
COLUMNS,
{set
{setdefining
definingthe
therow
rowheaders}
headers} ON
ONROWS
ROWS
FROM
FROM [cube
[cubename]
name]

15

Slide 15 of 27

Ver 1.0

Business Intelligence Fundamentals

Basic MDX Query Structure (Contd.)

SELECT
For
example, the following MDX query:
{ [Time].[YQM].[All YQM] } ON COLUMNS,
{ [Geography].[All Geography] } ON ROWS
FROM
[OLAP_GeoTime]

produces a table similar to the table shown

below:

16

Slide 16 of 27

Ver 1.0

Business Intelligence Fundamentals

Basic MDX Query Structure

(Contd.)

The WHERE clause is used to restrict the returned data to specific

dimension and member criteria. This is referred to as a slicer dimension. A
slicer dimension is expected to return data for a single member.
General form of the query with the addition of the WHERE clause:

SELECT
SELECT
{set
{setdefining
definingthe
thecolumn
column headers}
headers} ON
ONCOLUMNS,
COLUMNS,
{set
{setdefining
definingthe
therow
rowheaders}
headers} ON
ONROWS
ROWS
FROM
FROM[cube
[cubename]
name]
WHERE
WHERE(slicer)
(slicer)

17

Slide 17 of 27

Ver 1.0

Business Intelligence Fundamentals

Basic MDX Query Structure

(Contd.)

The following MDX query uses the WHERE clause to select the
measure:

SELECT
{ [Time].[YQM].[All YQM] } ON COLUMNS ,
{ [Geography].[All Geography] } ON ROWS
FROM
[OLAP_GeoTime]
WHERE
([Measures].[QuantitySUM])

18

and produces a table similar to the table shown below:

Slide 18 of 27

Ver 1.0

Business Intelligence Fundamentals

19

Named Sets and Calculated

Measures

You can use MDX to perform the following tasks:

specify a name for set expressions (named sets)
define formulas and treat each formula as a new member

Slide 19 of 27

Ver 1.0

Business Intelligence Fundamentals

Named Sets

20

A set can be a lengthy and complex declaration, and difficult to

follow or understand. A named set is an alias for a defined set. The
use of named sets helps to accomplish the following tasks:
clarify the logic by using an alias to simplify
the MDX code
potentially increase the efficiency of query execution because
the set is only executed once at the beginning, stored, and then
reused each time that it
is encountered
increase the ease of maintenance

Slide 20 of 27

Ver 1.0

Business Intelligence Fundamentals

Creating Named Sets for a Query

Use the WITH keyword as part of the SELECT statement to create a

named set as part of an MDX query.
General form for creating a named set using the WITH keyword:

WITH
WITHSET
SETset_name
set_nameAS
AS'set'
'set'
SELECT
SELECT
{set
{setdefining
definingthe
thecolumn
columnheaders}
headers} ON
ONCOLUMNS,
COLUMNS,
{set
{setdefining
definingthe
therow
rowheaders}
headers} ON
ONROWS
ROWS
FROM
FROM[cube
[cubename]
name]
WHERE
WHERE(slicer)
(slicer)

21

Slide 21 of 27

Ver 1.0

Business Intelligence Fundamentals

Creating Named Sets for a Query

The following MDX query uses the WITH keyword to create the
[North America] named set in the current query:

WITH SET [OLAP_CustProd].[North America] AS

'{[Demographic].[All CntryAgeGender].[Canada],
[Demographic].[All CntryAgeGender].[Canada].Children,
[Demographic].[All CntryAgeGender].[United States],
[Demographic].[All CntryAgeGender].[United States].Children
}'
SELECT
{ [Product].[All Product] } ON COLUMNS,
{ [OLAP_CustProd].[North America] } ON ROWS
FROM
[OLAP_CustProd]

22

Slide 22 of 27

Ver 1.0

Business Intelligence Fundamentals

Creating Named Sets for a Session

The CREATE SET statement creates a named set that exists for the
lifetime of the current SAS OLAP Server session. For example, the
following code creates the
[North America] named set:

CREATE SET [OLAP_CustProd].[North America] AS

'{[Demographic].[All CntryAgeGender].[Canada],
[Demographic].[All CntryAgeGender].[Canada].Children,
[Demographic].[All CntryAgeGender].[United States],
[Demographic].[All CntryAgeGender].[United States].Children
}'

The [North America]

named set is available
for use in future queries
in the current session.
23

Slide 23 of 27

Ver 1.0

Business Intelligence Fundamentals

Creating Named Sets Permanently

The CREATE GLOBAL SET statement creates a named set that

exists in the current and future SAS OLAP Server sessions. For
example, the following code creates the [North America] named set:

CREATE GLOBAL SET [OLAP_CustProd].[North America] AS

'{[Demographic].[All CntryAgeGender].[Canada],
[Demographic].[All CntryAgeGender].[Canada].Children,
[Demographic].[All CntryAgeGender].[United States],
[Demographic].[All CntryAgeGender].[United States].Children
}'

The [North America]

named set is available
for use in future queries
in the current session
and beyond.
24

Slide 24 of 27

Ver 1.0

Business Intelligence Fundamentals

Calculated Members

25

Calculated members have the following characteristics:

based on evaluated expressions in MDX
computed at run time
based on data that already exists
in the cube

Slide 25 of 27

Ver 1.0

Business Intelligence Fundamentals

Creating a Calculated Member for a Query

Similar to the way it is used in named sets, the WITH keyword is used to
describe calculated members.
General form for creating a calculated member using
the WITH keyword:

WITH
WITHMEMBER
MEMBER member-name
member-nameAS
AS 'value-expression'
'value-expression'
SELECT
SELECT
{set
{set defining
definingthe
thecolumn
columnheaders}
headers}ON
ONCOLUMNS,
COLUMNS,
{set
{set defining
definingthe
therow
rowheaders}
headers}ON
ONROWS
ROWS
FROM
FROM [cube
[cubename]
name]
WHERE
WHERE(slicer)
(slicer)

26

Slide 26 of 27

Ver 1.0

Business Intelligence Fundamentals

Creating a Calculated Member for

a Query (Contd.)

The following MDX query uses the WITH

keyword
WITH MEMBER [Measures].[QuantityYTD] AS
'(toSUM(YTD(
), QuantityYTD
Measures.[QuantitySUM])
create the
calculated)'measure
in
SELECT
CROSSJOIN
(
the current
query:

{[Time].[All Time].[1998].Children},
{ [Measures].[QuantitySUM],[Measures].[QuantityYTD]}
) ON COLUMNS ,
{[Product].[All Product].[Clothes & Shoes].Children}
ON ROWS
FROM
[OLAP_PTCStar]
27

Slide 27 of 27

Ver 1.0

Business Intelligence Fundamentals

Creating a Calculated Member for

a Session

The CREATE MEMBER statement creates a calculated member

that exists for the lifetime of the current SAS OLAP Server session.
For example, the following code creates the QuantityYTD member:

CREATE MEMBER
[OLAP_PTCStar].[Measures].[QuantityYTD]
AS '( SUM(YTD( ), Measures.[QuantitySUM]) )'

28

Slide 28 of 27

Ver 1.0

Business Intelligence Fundamentals

Creating a Calculated Member

Permanently

The CREATE GLOBAL MEMBER statement creates a calculated

member that exists in the current and future SAS OLAP Server
sessions. For example, the following code creates the QuantityYTD
member:

CREATE GLOBAL MEMBER

[OLAP_PTCStar].[Measures].[QuantityYTD]
AS '( SUM(YTD( ), Measures.[QuantitySUM]) )'

29

Slide 29 of 27

Ver 1.0