latrobe.edu.

au

Principles of Business Analytics

BUS5PB, Lecture 3
(S1 - 2024)

La Trobe University CRICOS Provider Code Number 00115M

 latrobe.edu.au

Recap
Week 1 Week 2

● Defining business analytics ● Digital Transformation

● Analytics environment: tools and data

● Types of analytics
● Anatomy of decisions
● Deriving values
● Analytics and strategy
● Business Analytics Lifecycle
● Analytics components
● Real-world success stories
● Organisational structures
● Challenges in analytics
● Analytics Roles
 Topic 3: First Tools in
Analytics

3
 latrobe.edu.au

Learning objectives
• First tools in Analytics
• Analytics dashboards
• Dimensional models
• Visualisation widgets
 latrobe.edu.au

First tools in Analytics (visualisation)

• The first tools in analytics are almost always focused on data visualisation and
the integration of such visualisations into interactive dashboards

• Microsoft Power BI, SAP Lumira, Tableau, Qlikview, Yellowfin BI,

Datawrapper etc…

• How about Microsoft Excel? Is it a spreadsheet or data visualisation tool?

• What are the capabilities you would expect of a data visualisation tool?
 latrobe.edu.au

Analytics dashboards
• A collection of visual (and text) widgets that represents all the data, insights and interaction on a
single screen

• Dashboards can also connect all types, formats and sources of data with different metrics and KPIs

• Who uses a dashboard?

• What is the first design principle when building a dashboard?

 latrobe.edu.au

Analytics dashboards

[Source: Dashboard Templates: Sales Dashboard (executiveknowledge.org)]

 latrobe.edu.au

Power BI in brief
• Gartner recognised Microsoft (Power BI) in the Leader Quadrant of “Analytics and
Business Intelligence Platforms” for 16 consecutive years.

• Let’s explore the Power BI website to understand this type of tools better

• As you will notice in the workshops, all analytics visualisation tools have similar
capabilities and features, and each year a new version is released with more features

• You should also appreciate that NOT all features can be discussed in class, which
means there is a substantial self-study component for designing and building the best
dashboards

• https://powerbi.microsoft.com/en-us/what-is-power-bi
 latrobe.edu.au

Dashboard best practices

• There are many guidelines and best practices by the different providers and vendors

• https://powerbi.microsoft.com/en-us/data-dashboards/

Follow Avoid

• A well defined objective driving the dashboard • The negation of the ‘Follow’ list and..
• Use the most up-to-date data • Too many widgets, cluttered visuals
• Group data logically • Incorrect or low-quality data
• Know your graphs and data type (line graph - temporal • Poorly formed graphs – missing labels, axes
data) • Trying to reach multiple audiences
• Know your audience – non/technical • Excess animation or expectation of user
• Maintain a hierarchy of information and insights interaction
• Consistent interface design
• Consistent use of colour • And many more…

• And many more…

 latrobe.edu.au

What is a Data Model ?

La Trobe University CRICOS Provider Code Number 00115M

 latrobe.edu.au

A first data model

• Interestingly, a fairly simple and intuitive data model underlies all of the
analytics tools discussed thus far - the dimensional model
• The dimensional model originated in the data warehousing era
• Formally defined as, a dimensional model is a database structure that is
optimised for online queries and data warehousing tools.
• It is comprised of "facts or measures” and "dimension or descriptors”
• Advantages: Intuitive to ‘business’ users, Fast query performance, Easily
extensible
 latrobe.edu.au

Dimensional model
Product
Product ID
Product Name
Product Type
Dimension Table

Customer Measures
Competitor
Customer ID Customer ID
Competitor ID
Country ID Product ID
Competitor Name
Country Age Group Competitor ID
Country ID (Customer Sentiment
Country Name Details) Revenue
Region Name Count

Fact Table
 latrobe.edu.au

Measures or facts
• A numeric variable expressing the KPIs, performance measurements, or any other
quantifiable information about an organisation.
• Measures are numerical variables and can have both discrete and continuous
values
• To ensure meaningful aggregations and accurate result, a numeric measure can be:
• Additive: can be summed up through all dimensions, i.e. can aggregate these
measures across any combination of dimensions without losing meaning or
accuracy, e.g. sales amount.
• Non-additive: cannot be summed across any dimension, e.g. unit price, ratios.
• Semi-additive: can be summed across some dimensions but not others, e.g.
account balance, inventory levels.
 latrobe.edu.au

Dimension or descriptors
• Dimensions describe the “who, what, where, when, how, and why” associated with the
measures.
• Customer – name, dob, address, preferences
• Dimensions are categorical variables, and take nominal, ordinal, binary or text types of
values
• A dimension is always used to “describe” (or spread out) an aggregated measure.
• A dimension “table” is a grouping of a similar set of dimensions (or attributes or columns),
like the “customer table”
• A hierarchy is a structure made up of two or more levels of related dimensions. A
dimension at an upper level of the hierarchy completely contains one or more dimensions
from the next lower level of the hierarchy.
• E.g. The ‘State’ dimension would contain one or more ‘Wine Region’ dimensions.
 latrobe.edu.au

Key elements of a dimensional model

So, now we have the key elements
of dimensional modelling …
Product Channel
• Measures (or Facts) — e.g. Sales
Performance
• Dimensions — e.g. Time, Product,
Market and Region
Customer Sales Store
• Keys — Surrogate key, Foreign
key, Composite key
Fact Table
• Attributes, Hierarchies, Schema Dimension
Table
Order Time

You will learn dimensional models more comprehensively in BUS5WB

 latrobe.edu.au

Data import..
• When a dataset (CSV file or database table) is imported into an Analytics tool
automatically creates a dimensional model based on the types and
occurrences of the data points in this dataset

• And when we click and drag fields on to the dashboarding canvas, we would
mainly visualising a numerical field with a categorical field, such as,

• Sales revenue by month

• Customer count by suburb

• Expenses by expense type

 latrobe.edu.au

Thinking in dimensions
• Measures are quantitative data items that give a numeric value to the data
(sentiment, count, revenue in this case).
• Dimensions add context to the measures and without context, measures are
meaningless numbers (age group, product type, competitor, region, country).
• We will discuss this further in the workshop.
 latrobe.edu.au

More dimensional models…

• Bringing the elements together in a star
schema.
• At the centre of the star is a fact table
containing the numeric measurements.
• The points and edges are dimension
tables that contain the descriptive context
of these measures.

Let's identify dimensions, facts, attributes,

primary keys, foreign keys in the example
schema.
 latrobe.edu.au

Snowflake Schema
• A snowflake schema
would contain a
normalised dimension
table to separately
represent a hierarchy.
• Normalisation is a
process of
decomposing the
dimension table to
reduce data
redundancy and
anomaly phenomenons.
 latrobe.edu.au

Why Dimensions ?
(and what does geometry have to do with this?)

La Trobe University CRICOS Provider Code Number 00115M

 latrobe.edu.au

An example – a winery
“We sell wine from different regions in various markets and measure our sales
performance over time.”

• We need to measure total sales performance –

• Let’s introduce a Time Dimension

 latrobe.edu.au

Winery – two dimensions

• “We sell wine from different regions in various markets and measure our sales
performance over time.”

• Introduce a Product (wine) Dimension

 latrobe.edu.au

Winery – three dimensions

• “We sell wine from different regions in various markets and measure our sales
performance over time.”

• And a Market Dimension

 latrobe.edu.au

Winery – four dimensions

• “We sell wine from different regions in various markets and measure our sales performance over time.”

• Wine region can be a dimension or a dimensional attribute.

• Not directly linked to sales performance, linked via product.

• More than three dimensions is not easily visualised but we can keep adding.
 latrobe.edu.au

Data Visualisation (just an introduction)

La Trobe University CRICOS Provider Code Number 00115M

 latrobe.edu.au

Visualisation widget vocabulary

• http://ft-interactive.github.io/visual-vocabulary/
• https://github.com/ft-interactive/chart-doctor/tree/master/visual-vocabulary
• https://public.tableau.com/views/VisualVocabulary/VisualVocabulary
 latrobe.edu.au

Visualisation widgets
• Also known as graphs, plots or charts
• A dashboard is composed of multiple widgets that come together to create
the interactive dashboard experience
• BUS5VA – Visual Analytics will delve deep into theory and practice of
visualisation for analytics.
• But here, we will explore some frequently used graphs as preparation for the
Analytics tools workshops
• Again, it should be emphasised that we cannot cover all types of graphs and
use cases, it is up to the student to engage in some self-study to further
develop these skills.
 latrobe.edu.au

Bar charts and stacked bar charts

• Use: visualising a measure (numerical) with dimensions (categorical), part-to-
whole relationships
• Suggestions: horizontal data label; consistent/contrast colours; ranking data
ascending/descending; use scale clear to view

[ Source: SAP Lumira (2018) ]

 latrobe.edu.au

Pie charts

• Used for: comparing percentage

values in proportion to the whole
• Suggestions: small number of slices
(<= 7); add up to 100%; easy to read
data labels; order slices based on size

[ Source: SAP Lumira (2018) ]

 latrobe.edu.au

Why not Pie Charts?

 latrobe.edu.au

Scatter plots

• Use: showing correlation between

two measures (numerical
variables), distribution trends,
similarities in datasets, identify
outliers
• Suggestions: use colours on dots
but not too many; keep the aspect
ratio square

[ Source: SAP Lumira (2018) ]

 latrobe.edu.au

Geo bubble charts

• Use: viewing a measure by Country, Region, or City; comparing measures
across different geographic areas
• Suggestions: show values on a map; animation; clear labels

[ Source: SAP Lumira (2018) ] [ Source: Tableau (2018) ]

 latrobe.edu.au

Tree maps and heat maps

• Use: comparing quantities by

categoricals using colour and size,
showing the distribution of measure
values
• Suggestions: only use when there
is visible difference in colour
intensities

[ Source: SAP Lumira (2018) ]

 latrobe.edu.au

Visual Elements

[ Source: https://clauswilke.com/dataviz/aesthetic-mapping.html ]
 latrobe.edu.au

Visual Elements
• Different data types require different elements

* The colours and shapes used for ordinal data should help readers immediately sort values from lowest to highest

• Example: https://public.tableau.com/app/profile/katelynschaub/viz/CuracaoSeaTurtles/CuracaoSeaTurtles-v3
 latrobe.edu.au

Dashboard: Top tips

• Know your audience

• Choose appropriate size and layout
• Use 5 views or fewer in dashboards
• Provide interactivity (not infographic)
• Most important view goes on the top or
top left
• Legends go near their views
• Tooltips where needed
• Avoid using multiple colour scheme on
a single dashboard (be consistent)

[ Source: https://www.tableau.com/solutions/gallery/investment-portfolio-summary ]
 latrobe.edu.au

References
• Stubbs, E. (2011). The Value of Business Analytics: Identifying the Path to Profitability (1st Edition).
• Provost, F. and Fawcett, T. (2013). Data Science for Business: What You Need to Know About Data Mining and Data-Analytic Thinking
(1st Edition).
• Sharda, R., Delen, D. and Turban, E. (2017). Business Intelligence, Analytics and Data Science: A Managerial Perspective (4th
Edition).
• Chapman, P., Clinton, J., Kerber, R., Khabaza, T., Reinartz, T., Shearer, C. and Wirth, R. (2000). CRISP-DM 1.0: Step-by-step data
mining guide. ftp://ftp.software.ibm.com/software/analytics/spss/support/Modeler/Documentation/14/UserManual/CRISP-DM.pdf, last
accessed on 18 March 2021.
• Coates, M. (2017). Designing a modern data warehouse + data lake.
https://www.sqlchick.com/s/DesigningAModernDWandDataLake_MelissaCoates.pdf, last accessed on 20 March 2021.
• BlueGranite (2020). Data lake in a modern data architecture. https://www.bluegranite.com/data-lakes-in-a-modern-data-architecture-
ebook, last accessed on 21 March 2021.
• SAP Lumira (2018). Data Visualization Handbook. https://cdn2.hubspot.net/hubfs/3901939/Analytics/Data-Discovery/SAP-Lumira-
Data-Visualization-Handbook.pdf, last accessed on 19 March 2021.
• Tableau (2018). Visual Analysis Best Practice: Simple techniques for making every data visualization useful and beautiful.
https://www.tableau.com/sites/default/files/media/whitepaper_visual-analysis-guidebook_0.pdf, last accessed on 19 March 2021.
• Financial Time (2020). Visual Vocabulary. https://github.com/ft-interactive/chart-doctor/tree/master/visual-vocabulary, last
accessed 23 March 2021.
 latrobe.edu.au

Workshop
[Analytics practice with MS Excel & Power Pivot]
 Thank you
latrobe.edu.au

La Trobe University
CRICOS Provider Code Number 00115M © Copyright La Trobe University 2018