01.

Why Learn Power BI ?

GIE_AXA_Internal Internal Document – June 2022

 Power BI is Microsoft tool
for Self-Service Business Intelligence:
Answering business questions by looking into data, all by yourself

It augments Excel functionalities, keeping a similar look & feel

GIE_AXA_Internal
 More than 115 million users
 Ranked #1 by Gartner
 AXA preferred choice for BI platform
 Embed Artificial Intelligence to easily get insights
 Work from your laptop or leverage cloud power
 Connect to any source of data

GIE_AXA_Internal
Power BI is an Extremely rich tool…
… but not all features are relevant for Audit Data Analytics

GIE_AXA_Internal
We focus on functionalities that extend what we do in Excel and Power
Query
In terms of Business Value this means:

Moving from: To:

• Executing pre-planned A/B Test Scenario • Exploratory Data Analysis (notably in scoping)
• Tables summarizing results • Leveraging visuals for insights & communication

Note: you could also do most of those things in Excel, just not as simple, and thus not as likely.
GIE_AXA_Internal
We focus on functionalities that extend what we do in Excel and Power
Query
In terms of Technical Value this means:

Moving from: To:

• ‘Small’ volume • Managing large volume of data
• ‘Flat’ dataset • Managing relationship in data
• ‘Simple’ calculation • Applying ‘complex’ calculations

Note: you could also do most of those things in Excel, just not as simple, and thus not as likely.
GIE_AXA_Internal
 Audit Data Analytics SharePoint
Demo

GIE_AXA_Internal
 02.
Build your first visuals

GIE_AXA_Internal Internal Document – June 2022

 FT Visual Vocabulary

GIE_AXA_Internal
Power BI Visuals

 A lot of visuals available by default… and many

more ‘custom’ ones to pick from the store.
 Most are straightforward to use, and
documentation is readily available inside the
tool.

GIE_AXA_Internal
Installation, Licenses and First connection
 PowerBI can be used either:
• Online in the cloud (« Power BI Service »); or
• As a software installation on your latptop (« Power BI Desktop »)
 In terms of license,
• The Desktop version is free (but you still need to login with a Microsoft account, eg. your PassAXA)
• The online version requires a license (Power BI ‘pro’, which costs less than 10$ a month);
it allows to share reports and datasets with other staff in the entity directly on the web.
• There also exists a ‘premium’ license (20$/month) to handle very large amount of data and leverage
advanced Artificial Intelligence functions.

 For this course, we will use ‘Power BI Desktop’ version; and if possible, a ‘pro’ license. You need to request
installation from your IT Helpdesk.

 Note that updates are released monthly, and thus to benefit from latest functionalities, you should regularly
update the installation.

 When you first launch the application, you are prompted to enter your PassAXA.

GIE_AXA_Internal
Power BI is organized in three layers of
functionalities

1 Power Query • Get & Prepare your data

• Add new columns via formulas

(price * qty)
2 Data/Relationship views • Manage relationship
(Link purchase to Invoice)

• Explore visually your data set

• Perform calculations
3 Report view
• Get Insights trough
off-the-shelve AI functions

GIE_AXA_Internal
Discovering the Interface
1. Login using your corporate email.
2. Name of your project.
3. Main Menu
4.‘Reports‘ tab where you can see
your created visualizations and
reports.
5.‘Data‘ tab to create and manage the
data you have imported in Power BI.
6.‘Model‘ tab will help you manage
the data models and relationships.
7.This is the area you can use to fill
in your visualizations and reports.
8.These are different chart types and
visualization you can use to design
your reports.
9.This is the ‘Field‘ section which
will show all the data imported, for
integration in the reports visual

GIE_AXA_Internal
 02.
Build your first visuals

Part 5: Other visuals & Conclusion

GIE_AXA_Internal Internal Document – June 2022

Other standard visuals
Integration with other MS/Azure services:
• Maps with Azure Bing to geolocate addresses
• Visio to format diagram based on data values
• Power Automate for RPA

AI assisted visuals (we ll see in next session):

• Decomposition Tree,
• Key Influencers,
• Q&A,

Python / R visuals
• Bring your own code or use pre-built ones.
Can easily be executed online, but on desktop
they require Python/R Install.
GIE_AXA_Internal
Custom Visuals from the Store
• They may provided by:
₋ Microsoft
₋ 3rd parties with certification
by Microsoft
₋ Fully provided by 3rd
parties

• Some are free, others need a

paid license

• Some may send data externally

• AXA admin may block access in

some entities.

• Some allow to ‘code’ your own

visuals with almost unlimited
possibilities
GIE_AXA_Internal
 Plot Limits
• PowerBI puts limit on the number of points displayed in each visual, so that the rendering remains fast.

• This can vary from 1000 to 150.000 points depending on the visual.

• In order for the insights to remain accurate, Power BI generally uses high density sampling strategies that
prioritises rendering of outliers, max/min, trends, etc.

• When reaching a limit or swicthing to sampling, Power BI display an alert icon on top of the visual (« ¡»)

Data point limits

High-Density Sampling
GIE_AXA_Internal
 To add for next update: zoom sliders directly in graph
Recap: what we covered in the session
Visuals Types Visuals Configuration

 Card, Multirow Card, Table, Matrix  Rename field names, as they appear in visuals
 Bar chart, line chart, area chart, ribbon chart  Prevent aggregation if not desired
 Pie chart and tree map  Manipulate [date] hierarchies
 Waterfall, Funnel  Apply conditional formatting
 Scatterplot and Bubble chart  Add Sparkline
 Others: Maps, Python, R, Customs, etc.  Copy/paste visuals, Undo actions
 Split in small multiples
 Split categories through legends or details
 Add tooltips
 Show items with no data
 Extract data / Export to Excel
 Edit the visual style (color, title, etc.)
 Add a play axis
Visual Types | Microsoft Docs  Drill-down (3 types)/up
Demo by Financial Times on Visual Vocabulary  Identify alert when max number of points reached
Demo on Outliers visuals  Add custom visuals from the store
Link to report built during part 2  Etc.
GIE_AXA_Internal
 03.
Get Insights by Analysing Data

GIE_AXA_Internal Internal Document – August 2022

Recap: what we covered in the session
List of Functionalities used
 Edit interactions
 Use Filter
 Use Slicer
 Multi select points/visuals
 Right click > Exclude some points
 Find anomalies (IA)
 Find Insights, Explain increase/decrease (IA)
 Use Clustering
 Get Insights on the whole dataset
 Use the Q&A AI visual
 Use the Decomposition Tree AI visual
 Use the Key Influencer visual
 Edit page size, lock to grid
 Setup Drill through
 Setup field parameters
 Setup visual tooltips

GIE_AXA_Internal
 04.
Export and Collaboration

GIE_AXA_Internal Internal Document – June 2022

Power Query online: DataFlows

GIE_AXA_Internal
Recap: what we covered in the session
List of Functionalities used
 Bookmarks
 Zoom is, show as table, show as datapoint
 Export data (desktop vs online)
 Copy/Paste data table from the model view
 Export as PDF
 Save File as PBIX or PBIT
 Publish Online
 Read vs Edit view
 ‘Get Insights’
 Share with others
 Add comments to a report
 Ability to Embed in Excel or PowerPoint
 Dataset Export (no limit)

GIE_AXA_Internal
 05.
Calculation and Modelling

Part 1: The Data Model

GIE_AXA_Internal Internal Document – June 2022

Data Models

Relational Model (database)

GIE_AXA_Internal
Data Models

Flat Model (Excel-like end user report)

GIE_AXA_Internal
Data Models
Star Model (Power BI and all Cube/Business Intelligence systems)

GIE_AXA_Internal
RelationShip and Filter Direction

GIE_AXA_Internal
Date Table

Date = CALENDARAUTO()

GIE_AXA_Internal
Recap: what we covered in the session
List of Functionalities used
 Create and Edit relationship between tables
 Change filter direction to both
 Principle of Star Data Model
 Create a date table, mark it as such and format it
 Create Display group and Hierarchies

GIE_AXA_Internal
 05.
Calculation and Modelling

Part 2: Making Calculation

GIE_AXA_Internal Internal Document – August 2022

Intro to DAX
Excel (with tables) Power BI

c =SUM(Table1[Units])

 Power BI uses the DAX language (Data Analytics eXpression)

 Basic usage is very similar to Excel
 Advanced usage can allow to make very complex calculation in a simple way (e.g. Time Intelligence)
 Although it may look simple, It is tricky to master and mistakes are easy to do!

Learn more: Use DAX in Power BI Desktop Data Analysis Expressions (DAX) Reference
GIE_AXA_Internal
Row Calculation via Calculated Columns
Unit Price Quantity Amount

1 5 5

Unit Price = DIVIDE(Purchases[Purchase Amount],Purchases[Quantity])

Expensive Item= IF(Purchases[Unit Price]>10000,TRUE,FALSE)

Requester and Approver = Purchases[Requester] & " - " & Purchases[Approver]

Cluster Refeference = RIGHT(Purchases[Vendor Name (clusters)],1)

GIE_AXA_Internal
Column calculation (aggregation)

… Qty

Total Purchased Amount = SUM(Purchases[Purchase Amount])

Total Number of purchase = COUNT(Invoices[PO ID])

GIE_AXA_Internal
Row + Column Calculations : Row Iteration

Unit Price Qty (Amount)

Total Purchased Amount =

SUMX ( Purchases, Purchases[Quantity] * Purchases[Unit Price] )

GIE_AXA_Internal
Multi-table calculation
Purchase Table Invoice Table
PO ID Purchased Amount Invoiced Amount PO ID Invoice Amount

PO-02 1000 € 900 € PO-02 500 €

PO-02 400 €

« 1 to many » relationship

Total Invoiced Amount = Sum(Invoices[INVOICE AMOUNT])

Total Related Invoiced Amount = CALCULATE (Sum(Invoices[INVOICE AMOUNT]))

Number of Invoices = COUNT(Invoices[Invoice ID])

Number of related invoices = CALCULATE(COUNT(Invoices[Invoice ID]))

GIE_AXA_Internal
Calculated Table

Invoices Monthly Summary =
SUMMARIZE (
    'Invoices',
    'Invoices'[Invoice Date].[Month],
    “Monthly Total", SUM ( 'Invoices'[INVOICE AMOUNT] )
)

2021 IT Purchases =
CALCULATETABLE (
    'Purchases',
    YEAR ( 'Purchases'[Purchase Date] = 2021 ),
    Purchases[CostCenter] = "IT"
)

Note: In both situation, we would obtain the same info. simply viaGIE_AXA_Internal
adding
N bothto the report a ‘table’ visual, with adequate filters and aggregations
Measures Implicit Measures Explicit Measures

Total Purchase Amount =  

SUM(Purchases[Purchase Amount])

20% of Total Amount Purchased in January only =

CALCULATE (
    SUM ( Purchases[Purchase Amount] ) * 0.2,
    Purchases[Purchase Date].[Month] = "JANUARY"
)
GIE_AXA_Internal
Recap: what we covered in the session
List of Functionalities used
 DAX syntax - Row/Column operations
 DAX CALCULATE function
 Calculated Columns
 Calculated Tables
 Measures
 Quick Measures
 Nature Language Processing for DAX
(to be released in 2023)
 Add Group/Binning measure
 “What if” Parameters

GIE_AXA_Internal
 06.
Capstone Project

GIE_AXA_Internal Internal Document – August 2022

GIE_AXA_Internal