Creating Text Columns for Data

Analysis

Andrew McSwiggan
BUSINESS INTELLIGENCE SPECIALIST
 Use a series of text manipulation
functions
Overview How to search through text strings to
extract specific values
Split the product description into
multiple columns
A firm foundation for product analysis
Add custom columns to your dimension
tables
Creating Columns
 Analysis Requirements

Season Product Item Product Type

Color Size
 Add Columns

Filters Slicers Display

 Creating Product Attributes

The Product Description is made up of a string of data elements

Autumn Winter Dress Classic White Large

We will use DAX to identify each element in the string

Season Product Type Color Size

Cut out each element from every record in the table

Create a new column attribute for each element

Save the new attributes and values in the product dimension table
Lists of values
 Creating Product Attributes

The Product Description is made up of a string of data elements

Autumn Winter Dress Classic White Large

We will use DAX to identify each element in the string

Season Product Type Color Size

Cut out each element from every record in the table

Create a new column attribute for each element

Save the new attributes and values in the product dimension table
Product descriptions
Look at the Data
How is it delimited?
Spaces!
Create a new description column
- Change the spaces
- To another value
- Not in the data
- Easy to see
Replace a String with SUBSTITUTE
 SUBSTITUTE(

<column_name>,

<old_text>,

<new_text>

<instance_num> )

SUBSTITUTE
Column Name of the column (or text string)

Search Text Text string to Search for

Replacement Text New text string

Instance Identify which occurrence to change

SUBSTITUTE(

<Product_Description>,

<“ “> ,

<“&”> ,

<1> )
Data Model
Row Context
Autumn Winter
Autumn&Winter

Spring Summer
Spring&Summer
 Create Column

Invokes Iterates each row Before users interact

“Row Context” with the model
Row Context.
 Create a DAX Expression

More beneath the Create new column Expression populates

surface each row
NESTING

SUBSTITUTE(

<SUBSTITUTE([Product_Description]," ","&",1)>,

<“ “> ,

<“-”> ,
<> )
Identify Text Positions with SEARCH & FIND
 Locate the Position

FIND SEARCH
Case sensitive Not case sensitive
 SEARCH(

“-”,

[Product Base],

1 ,

-1 )

SEARCH Returns Character Position

Find Text String to find

Within Text Column name or Expression

Start Num Character position in text to start from default = 1

Not Found Value set to -1

 Locate the Text Positions

Season - Product - Product Type - Color - Size

14 22 30 35
Cut Out Text Parts with LEFT MID & RIGHT
 Function Collection

Cutting functions Educational

Not official term
 Cut Out the Attributes

LEFT MID RIGHT

Season - Product - Product Type - Color - Size

14 22 30 35
 New Columns

Season Product Item Product Type

Color Size
 LEFT (

[Product Base]

, [Season Gap] -1 )

LEFT Returns a string to the left of another string

Source Column The column that contains the string to evaluate

Number of Characters The number of characters from the left

 MID (

, [Product Base]

, [Season Gap] +1

, ([Product Gap] - ([Season Gap] ) -1 )

MID Cuts out a sub string from the middle of a string

Column A column that contains the block of source text

Start Num The position in the middle of the string to start from

Length The length of the text to cut out - number of characters

 RIGHT (

, [Product Base]

, LEN([Product Base]) - [Post Color Gap] )

RIGHT Returns a string to the right of another string

Source Column The column that contains the string to evaluate

Number of Characters The number of characters from the right

 Cutting Functions

LEFT MID RIGHT

Data Cleansing with SUBSTITUTE
Messy Text
 Clean the Text String

SUBSTITUTE

Light Blue LightBlue

Light Green
LightGreen
Capitalising Text with UPPER & REPLACE
 Product
Title or - Lower case
Relevant
Other Attributes
Graphic - First letter is Capital
 CAPITALISE

UPPER ( <column> )

UPPER ( LEFT ( [Product],1 ) )

REPLACE ( [Product], 1 , 1 , UPPER(LEFT([Product],1)) )

 LOOKUP Values

Lookup

Dimension Table Fact Table

 Logic
- IF

Manipulate Text with DAX

Summary - SUBSTITIUTE
- SEARCH
- FIND

Cutout Text
- LEFT
- MID
- RIGHT

Lookup Values
- RELATED
- RELATEDTABLE
Join Text Strings with COMBINEVALUES
 Combine Text Strings

[Product Type] & "/" & [Product]

CONCATENATE( CONCATENATE( [Product Type] , "/" ) , [Product] )

COMBINEVALUES( "/" , [Product Type] , [Product] )

Lookup Values with RELATED and
RELATEDTABLE
 Lookup Functions

RELATED RELATEDTABLE
Single value Table of values
 RELATED ( <table name column name> )

RELATED

Table Name The name of the related table

Column name The name of the column

 Compare Dates

Registration Date Sales Date

Fact Table Row Context
Fact Table Row Context
Fact Table Row Context
Data Model
 Customers First Day Sales

First Day Other Days

Sales on date of registration Sales on a later date
Denormalisation
Create an Aggregated Column with
RELATEDTABLE
 Multi Sale

Single Sale Multiple Sales

Sale Count = 1 Sale Count > 1
 Multi Sale

Single Sale Multiple Sales

Customer has only purchased 1 Customer has purchased more
item than 1 item
Lookup Values
 Key Learning Points

Create columns Row Context

Demo
Create two columns
- Customer table

Sale Count
Multi Sale
 SUMX(

RELATEDTABLE ( <Table name> )

, <Column Name> )

RELATEDTABLE

SUMX Aggregation Function that gives a total

Table Name The name of the related table

Column name The name of the column

Row Context