DATA4000 Introduction to

Business Analytics

What is Business Analytics?

Workshop 1

Copyright Notice

COPYRIGHTCOMMONWEALTH OF AUSTRALIA

Copyright Regulations 1969WARNING

This material has been reproduced and communicated to you by or on behalf of

Kaplan Higher Education pursuant to Part VB of the Copyright Act 1968 (the Act). The material in this communication may be subject to copyright under the Act. Any

further reproduction or communication of this material by you may be the subject of

copyright protection under the Act.

Do not remove this notice

This Topic’s Big Idea

Marr, B 2015, ‘Big Data: 20 Mind-Boggling Facts Everyone Must Read’, Forbes, viewed 7 March 2017,

www.forbes.com/sites/bernardmarr/2015/09/30/big -data-20-mind-boggling-facts-everyone-m ust-

read/#6a4477ea17b1

“Artificial intelligence, however you want to

define it, that's everything. There will be more

changes in the next five to seven years than

we've seen in the last 30. It will impact every

business. Data is the new gold. It's the new

oil. It's the new plastics.”

Marc Cuban

"By the year 2020, about 1.7 megabytes of

new information will be created every second

for every human being on the planet… By then,

our accumulated digital universe of data will

grow to…around 44 zettabytes, or 44 trillion

gigabytes.”

Bernard Marr

Learning Objectives

In this workshop, we will:

1. Define business analytics and review applications

2. Distinguish between operational and discovery

analytics

3. Review and classify the different levels of

analytics with reference to real-world examples

4. Discuss traditional statistical approaches to

business problem-solving and contrast these to

new analytics, artificial intelligence, and machine

learning methods

General Expectations for DATA4000

• Participation

• Private study

• Discussion of

Assessments

• Housekeeping

Any Questions?

Assessments

• Assessment 1: Individual Case Study (2000

words)

• Assessment 2: Data Management, Analysis

and Visualisation Software Project (1000 words

+ visuals)

• Group Assessment 3: Group Report and

Presentation: My Health Record (10 slides +

1800 words)

What is Analytics?

• Analytics is a broad term covering both statistical analysis,

data mining and data-driven techniques and algorithms

• Analytics takes advantage of the variety and large volume

of data (real-time and past) available today

• The main aim of analytics is to gain useful information

(using analyses and algorithms) and business insights from

the data

This Photo by Unknown Author is licensed under CC BY

Activity 1: Data Journalism

Watch the video “The Age of Insight: Telling Stories with

Data” and answer the questions.https://www.youtube.com/watch?v=TA_tNh0LMEs

Discussion questions:

1. How does data journalism differ from traditional journalism?

2. What benefits and costs for journalists and society are associated with

data journalism?

Business Analytics

• Includes many qualitative and quantitative techniques

• Always starts with a business question/problem

• More than statistical analysis and “data-driven” decision making

• Approaches and algorithms applied to the data are flexible and run without

being directed too closely by hypotheses – the data “speaks for itself”

Definition: Using data-based algorithms, analysis and visualisation to guide business decisions and actions

This Photo by Unknown Author is licensed under CC BY-SA

Activity 2: Analytics and ElectionsHow the Trump Campaign used Big Data

Cambridge Analytica “harvested private information from the Facebook profiles of

more than 50 million users without their permission, according to former Cambridge

employees, associates and documents, making it one of the largest data leaks in

the social network’s history. The breach allowed the company to exploit the private

social media activity of a huge swath of the American electorate, developing techniques that underpinned its work on President Trump’s campaign in 2016.”

The demographic information of people who seemed to be pro- Republican

on social media was matched to voter information in the Republican

Party’s databases. Trump’s campaign analysts could then work out exactly

where to target their message, for the minimum expense.

Discussion questions:

1. How were data sources used to gain political advantage?

2. Is this ethical? W hy or why not?

Readings: https://www.nytimes.com/2018/03/17/us/politics/cambridge-analytica-trump-campaign.html

https://www.theguardian.com/uk-news/2018/mar/23/leaked-cambridge-analyticas-blueprint-for-trump-victory

Business Analytics Cont.

▪ Operational analytics drive day-

to-day operations and decisions

Includes diagnostics, automation

of billing, supply chain tasks,

manufacturing, etc.

▪ Discovery (Innovation) analytics

creatively drive new solutions,

products and services, e.g. new

apps, robots, suggestions for

customer purchasesThis Photo by Unknown Author is licensed under CC BY-NC-SA

Operational Analytics Case Study

In response to day-to-day fluctuations in oil prices (per barrel), a multi-

national company asked Deloitte to estimate how much the company

could save in costs if they used robotic process automation (RPA) rather

than manual operations.

https://www2.deloitte.com/us/en/pages/deloitte-

analytics/articles/business-analytics-case-studies.html

• A Deloitte project team worked out the

requirements for the company.

• Software called UIPath was developed

to carry out the automations.

• This is just the first step and so far has

saved the client 1,700 hours or labour per year (mainly related to supply chain

tasks).

Robotic process automation in oil and gas

Activity 3: Chatbots Case Study

1. Two Artificial Intelligence (AI) Chatbots talk and argue with

each other

2. Chatbot sitcom

Discussion questions: 1. What type of analytics are these chatbots an example

of and why?

2. Do the characters sound rational?

3. Do you think these chatbots could sit in on a

corporate board and solve a serious problem? Consider: https://sloanreview.mit.edu/article/ai-in-the-boardroom-the-next-realm-

of-corporate-governance/

Watch the below two videos and answer the questions below:

Different Types of Analytics

Descriptive analytics

Predictive analytics

Prescriptive analytics

Automation

Visualisations and summary statistics on

past data, e.g. last year’s median house

price, average monthly profit, etc.

Model based on past data and used to predict

a future outcome, e.g. Regression model for

predicting heart disease

Simulation, AI or optimisation algorithm which

suggests that one outcome is a better choice

than another, e.g. transport optimisation

Putting actions in the hands of computers or

robots. Smart home devices, e.g. Google

Home

Activity 4: Different Levels of

Analytics Video

Descriptive analytics

Predictive analytics

Prescriptive analytics

Automation

Discussion questions:

1. How do prescriptive analytics

work?

2. How do they differ from

descriptive analytics?

Descriptive Analytics Examples

Visualisations and summary statistics on past data, e.g. median

apartment & house prices for each Melbourne on March 2018

• Does not generally explain why an event happened

• Covers descriptive statistics and basic data mining techniques

https://www.domain.com.au/product/house-price-report-march-2018/

Predictive Analytics Examples

Model based on past data and used to predict a future outcome,

e.g. time series model projections of the average size of a new

solar photovoltaic cells (PV) system during 2018 to 2020.

This Photo by Unknown Author is licensed under CC BY-SA

http://www.cleanenergyregulator.gov.au/ Docum entAssets/Documents/Modelling%20report%20by%20Jacobs%

20-%20January%202018.pdf

Predicted

average PV

system size

Prescriptive Analytics ExamplesSimulation, AI or optimisation algorithm which suggests that

one outcome is a better choice than another, e.g. Deep Neural

networks for YouTube recommendations

Source: https://storage.googleapis.com/pub-tools-public-publication-data/pdf/45530.pdf

This Photo by Unknown Author is licensed under CC BY-SA

Automation Examples

Putting actions in the hands of computers/robots

Walking robots, which is prescriptive analytics in action

https://robohub.org/envisioning-the-future-of-robotics/

Driverless CarsHolistic Example of all Types of Analytics

Uses all levels of analytics (research)

➢ Descriptive – to assess the environment

➢ Predictive analytics – used to predict the next step

➢ Prescriptive – what should the car do next

➢ Automation – taking humans out of the equation See http://dataconomy.com/2015/12/how-data-science-is-dri vi ng-the-dri verless-car/

This Photo by Unknown Author is licensed under CC BY-NC-ND

Activity 5: Different Levels of

Analytics

Form a group and classify which type of analytics these items belong to:

1. A chart with median income of different professions

2. Driverless car

3. Robotic vacuum

4. Chatbot

5. Finance forecasting equation

6. Machine learning algorithm for customer sentiment analysis

7. Artificial intelligence algorithm for distinguishing weeds from tomatoes

8. Graph of the share price for Qantas over the past year

The Evolution of Analytics

Diagnostic analytics sits between descriptive and

predictive analytics. It examines data to explain why an

event occurred.

There is also alignment

with concepts such as:

* Hindsight

(looking to the past)

* Insight

(deeper understanding)

* Foresight

(Seeing to the future)

Source: www.tdwi.org

• Data driven solutions are becoming a reality

– Large data sets with a number of business variables can be

processed.

– In the past data analysis was limited to estimating parameters of a

known statistical distribution and testing that the solution conformed

(or not) to a hypothesis.

• For example

– They proposed a hypothesis e.g. “Class attendance positively affects

grades”

– They tested this hypothesis by applying a statistical test using

student data

Solving Business Problems

Traditional Statistical Method

General method

• A model, along with very specific hypotheses were created and tested

• Small samples with limited descriptive variables (characteristics) were

used because:

– Computing resources for multiple variables or large samples was

limited – small samples saved time and money

– Fitting multiple variables to a known distribution was intractable.

– The selection of the sample itself would have limited the possible

outcomes to some extent

Today’s Brave New World of

Artificial Intelligence

Why do we now use machine learning

(ML)?

• Availability of computing resources

• ML starts with minimal assumptions

• Can average over a variable

if you want to simplify analysis– This allows for streaming analytics

Source: https://www.nature.com/articles/nmeth.4642.pdf

Customer Churn – Statistics gives a coarse

solution. ML provides a finer and more nuanced

solution space

Old and New Approaches to Gene

DiscoveryBusiness problem: linking gene patterns to a particular disease

Statistical Method

Output is a test statistic

(p value compared to

test cut-off value)

• Test multiple hypotheses based on

the mean (average) expression of

certain genes

• The assumptions are based on

well known probability distributions

and pre-existing knowledge

• Genes are identified based on assumptions related to distributions

ML Method

• Try several algorithms and simulations and evaluate them using a range of

methods– No knowledge of gene sequencing is

required

• Genes are identified based on a non-

probabilistic method (feature

importance)

• Output is directly related to the gene

expressionSource: https://www.nature.com/articles/nmeth.4642.pdf

Activity 6: Using Data

Food myths (stories about food that are

not true). The sale of food is driven by

many myths rather than big data

analytics. Consider the following food

myth: Consuming several soft drinks per day is not harmful to adult health

W hat data do you think you would need,

and how would you attempt to

prove/disprove this myth:

1. Using statistics?

2. Using machine learning and big data?

This Photo by Unknown Author is licensed under CC BY-ND

Final Case Study

Business Problem:

How to deal with

Metabolic Syndrome

This Photo by Unknown Author is licensed under CC BY-NC-ND

Business Problem: How to Deal

With Metabolic Syndrome

Metabolic Syndrome covers a number of disorders with

symptoms ranging from high blood pressure, tendency to

gain weight, cholesterol issues and high glucose levels.

Source: https://gigaom.com/2012/11/20/how-aetna-is-usi ng-big-data-to-improve-patient-health/

Aetna is a healthcare insurance company

Aetna decided to solve the problem of Metabolic Syndrome

amongst its 18.2 million members.

How did Aetna approach this problem?

Short answer: as “scientists”. In business increasingly we will

have to operate as scientists (and “anecdotal” experience in

business will not count as much).

Source: https://gigaom.com/2012/11/20/how-aetna-is-usi ng-big-data-to-improve-patient-health/

Data available from members:

– Conditions treated (or billed for)

– Prescriptions filled

– The types of treatments doctors prescribe

• Aetna launched “Innovation Labs” in 2012.

Business Problem: How to Deal

With Metabolic Syndrome Cont.

Innovation Labs wanted to:

• Improve patient safety – screening for ineffective medication and

poor (harmful) combinations

• Customise patient care – allow doctors access to best practice,

latest treatments, up-to-date technology

• Increase patient engagement – change behaviour to healthy life

styles

Source: https://gigaom.com/2012/11/20/how-aetna-is-usi ng-big-data-to-improve-patient-health/

Aims

Business Problem: How to Deal

With Metabolic Syndrome Cont.

Aetna: Data Analytics Problem How to Deal With Metabolic Syndrome

Big Data Methods

– 600,000 lab results related to

screening were assessed.

(1.3 terabytes of data)

Analytics was used to:

– Personalise patient risk

– Personalise treatment

Outcomes:

– 90% of patients who didn’t

have a previous visit with their

doctor would benefit from a

screening

– 60% would benefit from

improving their adherence to

their medicine regimen

– Doctors now asked to focus

on intervention programs

Source: https://gigaom.com/2012/11/20/how-aetna-is-usi ng-big-data-to-improve-patient-health/

Next Week: Making the Move to Analytics

• Emerging job roles

• Individual adaptation in an evolving business world

• Case studies of professionals making the move to analytics

• New concepts to discuss and explore