CH5: business intelligence & data analytics
Introduction
The digital (r)evolution
- Pollev: which one is digital clock?: not left, uses analog method, something
that rotates => no digital technology
=> Right one has screen, it uses bit sequences to keep track of the time
- Most important technologies 20st and 21st century?: computer can be a good
answer, was invented by multiple people who build on it
=> No single inventor, so more like an evolution than a revolution (IBM, Alan
Turing, Charles Babbage…)
- Many steps, Big Data first, now AI => Importance of unstructured data (not
easily stored in tables)
- AI: Needs massive amounts of data to be trained, with big data that was made
possible
- Driving force?: money(work in more efficient matter, develop new markets)
=> Lot of investments help breakthroughs in technology
Hype cycle:
- Which are emerging technologies that are important for businesses?
- AI: going a bit down from the peak, plenty of people are getting critical of it =>
Should we invest this much, it is not solving everything
Impact on businesses
- Offices back then: not a single computer, only phone and typing machines
- Extensive investments in business infrastructure
- To support virtually every business function (marketing, manufacturing
- Data is new oil, collected as much as possible, used to innovate and improve
the business
=> Be more efficient, make more money
Data
= Everything that we can capture, store and describe as a sequence of bits
=> Music, videos, photo’s, text…
Information
- Difference with data: When the data has value/use for the user, data for one
person may be information for another person => subjective
- Utility: When we say value in economics, often utility is implied
- Knowledge: different than information, more useful, things in our minds
- External: data => information => internal: knowledge => wisdom
- From left to right more utility and aggregation & processing
1
,From data to information
- You need something that processes the
data into something useful (program)
- Program: algorithm with a finite sequence of
steps to transform an input into an output
- Use the information to create knowledge
and wisdom
Example bankloan
- Lots of data needed: loan history, income, stability, family, …
- Need to be collected and made into information
Sources of data
- Internal data: OLTP systems, (online transaction processing) en ERP system
(enterprise resource planning), CRM system (customer relation management)
- External data: Market trends, industry data, competitor information => You
can buy data about these topics
Different types of data (sl18!)
- Tabular: if you can store it in a table such as excel
- Non-tabular: images, video, book
- Before: storing data in tables, use te least amount of data possible
- Now: lot of unstructured data
- Structured vs unstructured
- Real time data vs historical data
- Experimental vs observational data
- Big data vs small data
Business intelligence and analytics
=> Measuring data to be able to understand and improve processes
=> Use to be hard to obtain, now much easier
=> Opportunities
- Getting to know what you need to know
- By analyzing data, applying data analytics
- Transforming data into information and knowledge
- Business: improve efficiency of business decision making
=> BI and analytics
- Efficiency: top line and bottom line
- Decisions: everything we do that we can control and change
2
, BI and analytics
- Data science for business
- Science vs business, theory vs application, statistics vs engineering
- Oriented towards providing decision support
- Method: Essentially any method that transforms data into information, insight,
knowledge… => Y = f(X)
- From statistics to machine learning (exceptionally powerful)
Machine learning for business decision-making
=> Is about many different methods
Business decision making (sl24 for example
- ML Mostly suitable for operational decisions
- Strategy level: high level, what products will we sell, what are the technical
evolutions, often very high impact but not that concrete
- Tactical level: mid level, managers, translate strategy into operations
- Operational level: Low level, short term, high frequency, daily activities, have
immediate effect
- The operational activities can often be automated through IS, because there
is lots of data available (such as customer data) and the decision complexity
is rather low
Machine learning
- Algorithm: sequence of steps to get from a problem to a solution
- Computer algorithm: can be implemented on a computer, uses input to get
output, uses data as input => Y = f(X)
- Programming: Algorithm => program, complex operational task?
- Complex to program
- e.g. Identify fraudulent transactions, very hard to detect one right
because there are so many tasks
- Can a computer learn to solve complex operational tasks, rather than a
human having to program it => It programs itself
- Can a computer learn f rather than a human having to program it
=> Machine learning! (= subfield in larger field of AI)
- Difference ML and machine reasoning:
- Learning: once learned or acquired: knowledge can be directly applied,
no further think required: fast thinking (reflexes
- Reasoning: Refers to solving a new problem: no existing solution
method available, involves active thinking: slow thinking
=> LLM such as chatGPT aren’t so good at solving riddles, they have
been trained through learning from processing huge amounts of data (it
can predict the next word in a sequence through learning)
=> haven’t been built for reasoning
3
Introduction
The digital (r)evolution
- Pollev: which one is digital clock?: not left, uses analog method, something
that rotates => no digital technology
=> Right one has screen, it uses bit sequences to keep track of the time
- Most important technologies 20st and 21st century?: computer can be a good
answer, was invented by multiple people who build on it
=> No single inventor, so more like an evolution than a revolution (IBM, Alan
Turing, Charles Babbage…)
- Many steps, Big Data first, now AI => Importance of unstructured data (not
easily stored in tables)
- AI: Needs massive amounts of data to be trained, with big data that was made
possible
- Driving force?: money(work in more efficient matter, develop new markets)
=> Lot of investments help breakthroughs in technology
Hype cycle:
- Which are emerging technologies that are important for businesses?
- AI: going a bit down from the peak, plenty of people are getting critical of it =>
Should we invest this much, it is not solving everything
Impact on businesses
- Offices back then: not a single computer, only phone and typing machines
- Extensive investments in business infrastructure
- To support virtually every business function (marketing, manufacturing
- Data is new oil, collected as much as possible, used to innovate and improve
the business
=> Be more efficient, make more money
Data
= Everything that we can capture, store and describe as a sequence of bits
=> Music, videos, photo’s, text…
Information
- Difference with data: When the data has value/use for the user, data for one
person may be information for another person => subjective
- Utility: When we say value in economics, often utility is implied
- Knowledge: different than information, more useful, things in our minds
- External: data => information => internal: knowledge => wisdom
- From left to right more utility and aggregation & processing
1
,From data to information
- You need something that processes the
data into something useful (program)
- Program: algorithm with a finite sequence of
steps to transform an input into an output
- Use the information to create knowledge
and wisdom
Example bankloan
- Lots of data needed: loan history, income, stability, family, …
- Need to be collected and made into information
Sources of data
- Internal data: OLTP systems, (online transaction processing) en ERP system
(enterprise resource planning), CRM system (customer relation management)
- External data: Market trends, industry data, competitor information => You
can buy data about these topics
Different types of data (sl18!)
- Tabular: if you can store it in a table such as excel
- Non-tabular: images, video, book
- Before: storing data in tables, use te least amount of data possible
- Now: lot of unstructured data
- Structured vs unstructured
- Real time data vs historical data
- Experimental vs observational data
- Big data vs small data
Business intelligence and analytics
=> Measuring data to be able to understand and improve processes
=> Use to be hard to obtain, now much easier
=> Opportunities
- Getting to know what you need to know
- By analyzing data, applying data analytics
- Transforming data into information and knowledge
- Business: improve efficiency of business decision making
=> BI and analytics
- Efficiency: top line and bottom line
- Decisions: everything we do that we can control and change
2
, BI and analytics
- Data science for business
- Science vs business, theory vs application, statistics vs engineering
- Oriented towards providing decision support
- Method: Essentially any method that transforms data into information, insight,
knowledge… => Y = f(X)
- From statistics to machine learning (exceptionally powerful)
Machine learning for business decision-making
=> Is about many different methods
Business decision making (sl24 for example
- ML Mostly suitable for operational decisions
- Strategy level: high level, what products will we sell, what are the technical
evolutions, often very high impact but not that concrete
- Tactical level: mid level, managers, translate strategy into operations
- Operational level: Low level, short term, high frequency, daily activities, have
immediate effect
- The operational activities can often be automated through IS, because there
is lots of data available (such as customer data) and the decision complexity
is rather low
Machine learning
- Algorithm: sequence of steps to get from a problem to a solution
- Computer algorithm: can be implemented on a computer, uses input to get
output, uses data as input => Y = f(X)
- Programming: Algorithm => program, complex operational task?
- Complex to program
- e.g. Identify fraudulent transactions, very hard to detect one right
because there are so many tasks
- Can a computer learn to solve complex operational tasks, rather than a
human having to program it => It programs itself
- Can a computer learn f rather than a human having to program it
=> Machine learning! (= subfield in larger field of AI)
- Difference ML and machine reasoning:
- Learning: once learned or acquired: knowledge can be directly applied,
no further think required: fast thinking (reflexes
- Reasoning: Refers to solving a new problem: no existing solution
method available, involves active thinking: slow thinking
=> LLM such as chatGPT aren’t so good at solving riddles, they have
been trained through learning from processing huge amounts of data (it
can predict the next word in a sequence through learning)
=> haven’t been built for reasoning
3
