Week 1
Videos 1a
Problem area: Defining a problem
Problem areas can be found through observation/curiosity, improving the status quo or
theoretical/conceptual issues to understand phenomena and a potential to become a researchable
project, because it is ongoing now.
A good research problem is:
• Developed from a sound theoretical base
• Of interest to sponsor and researcher
• Well defined
• Clear budget
• Scientific contribution > literature review needed
1.1: Selling data: Business model issues
Data which is collected in your business, e.g. through sensors, apps or websites, can be valuable for
other parties as well.
In order to sell this data, you need to create a business model. For this, you need to understand what
data people want, and who wants to buy it. Next, you have to decide on a price. Since data is already
there, the costs will be low. Lastly think of how you will send data to your buyers. Nowadays, data
marketplaces are emerging.
There are restrictions on selling data however. Personal data often should be anonymized. But even
if this is done, people could ‘guess’ who the data belonged to. There are also legal/ethical issues. You
need user permission for personal data sharing. You need to know which questions should be asked.
1.2: Selling Data: Technology issues
Personal data can contain personal sensitive information, which people do not want to be shared.
Simply removing info such as name, address, is not enough. Since quasi-identifiers combined can still
allow people to be identified. Therefore, there are anonymization measures in place:
• Data landscape analysis here you become aware of the de-anonimisation risk in your data
set and consider how an adversary would deanonymize individuals. You need to answer the
following questions:
o What info in your data set can be obtained by outsiders
o How much effort does it take to obtain the info
o How severe are the consequences of deanonymization
• Threat analysis consider the likelihood of deanonimisation. Various tools for this exist.
• Anonymization measures these are used to distort your data set. While they protect privacy,
they also distort its utility, so you need to decide how distorted the data set should be. You
can reach how much to distort based on 3 factors:
o The effort (cost, time and know-how) to deanonymize
o The likelihood of a deanonymization
o The severity of the consequences of a deanonymization in your data set
,1.3: Selling data: Ethical and Regulatory Issues
The first and most important regulation is the General Data Protection Regulation (GDPR). These
rules apply in almost the entire EU. The GDPR with personal data only.
Personal data: all data linked, directly or indirectly, to an individual called a data subject. Examples
are your address, a like on facebook and your IP.
If data cannot be qualified as personal, you do not have to follow the GDPR rules.
Processing of personal data: all activities that involve the use of personal data, includes collection,
consultation and anonymization. If anonymized correctly, the GDPR rules do not have to be followed.
In all data activities there are 2 entities, the data owner and the data controller: who processes the
data. The data controller might elocate some tasks to the processor, who has fewer responsibilities.
If there is more than one who owns the data, they are called joint controllers. To clarify who is
responsible for what, they need to sign a contract.
Personal data are not a commodity. The data subjects always have fundamental rights related to
them, even if they give it away for using a service.
There are also essential ethical principles, used to interpret the law. The European Data Protection
Supervising authority (EDPS) has developed ethical guidelines for this. One important distinction
here is the right of data subjects to not be subject to a decision based solely on automated
processing.
Buying data: Business Model Issues
A data driven business model (DDBMI) has the following characteristics:
• Data is used as a key resource
• Data analytics methods are used
• Data is part of the value proposition
How can data help your business model? 5 fundamental questions need to be answered:
• What do we want to achieve by using data?
o Use data for strategic decision making, improving processes
o Enrich core products and services with information
o Sell information offerings
• What is our desired offering?
• What data do we require and how are we going to acquire it?
o What data do we already get from existing processes? What do we get from our
products and customers and what from data providers/market places.
• In what ways are we going to process and apply this data?
o Sometimes requires expert knowledge
• How are we going to monetize it?
o Direct monetization use a subscription model, consumption based (pay-per-use) and
value-based model
o Indirect monetization, increase price of core product/service, increase sales volume
of core product or service and increase quality or reduce costs
,Buying data: Technology issues
Multi-party Computation (MPC) allows business owners to securely compute the sum of their data,
without revealing their separate data.
Issues with MPC however are that even though there might be an economic benefit, sharing the data
might be too much of a risk, due to violation of the GDPR or revealing inside information of your
business.
The aim of MPC is to enable a group of participants to jointly compute an agreed upon function,
without revealing the private input to the other participants. There is no need for a trusted 3rd party.
The most popular concept underlying MPC is secret sharing. Secret sharing can be divided into 3
steps:
• First secret shares are created
• Shares are exchanged
• The computation is done, the sums of the secrets are added
Week 1b
Why do managers need research?
(business) research is an organized, systematic, data-driven, critical, objective, scientific inquiry or
investigation into a specific problem.
Basic/fundamental research Applied
Generate body of knowledge on class of Solve specific business problems faced by
problems someone in work setting, demanding timely
solution
New contribution to scientific research Apply existing theory to solve a problem
What makes research scientific?
• Scientific statements can be verified (empirically testable!)
• Purposiveness (aim, focus)
• Testability and replicability
• Precision and confidence
• Objectivity
• Generalizability
• Parsimony (you want things found as simple as possible, but not simpler than that)
The empirical cycle
The principle of verification = you should find observations in support of your claim and start
reasoning from this empirical finding (i.e. to find ‘truth’) through induction.
The principle of falsification = a claim remains ‘true’ until proven otherwise through deduction (karl
popper).
Empirical cycle: Observation > induction > deduction > testing > evaluation >
, The research process in detail:
• Identify (broad) research problem
• Literature review
• Research objectives, questions, hypotheses
• Select research approach/research design
• Create plan for research
o Sample, instruments, data collection approach
• Collect and analyze data
• Interpret data and evaluate validity
• Communicate findings/write report
Research objective question part 1
Defining the research objective:
• Acquiring new knowledge
o To contribute to solving a problem
o To increase existing knowledge on a specific area
• The goal of research, when it is achieved
Defining the research question:
• The goal in research
• What should you know to realize the research objective?
o Should be logically deduced from the research objective
• More specific than the research objective
o How frequent is…; what is the relation between..
• Should be aimed at knowledge, not strategy/policy
• Should be empirical (not normative)
Videos 1a
Problem area: Defining a problem
Problem areas can be found through observation/curiosity, improving the status quo or
theoretical/conceptual issues to understand phenomena and a potential to become a researchable
project, because it is ongoing now.
A good research problem is:
• Developed from a sound theoretical base
• Of interest to sponsor and researcher
• Well defined
• Clear budget
• Scientific contribution > literature review needed
1.1: Selling data: Business model issues
Data which is collected in your business, e.g. through sensors, apps or websites, can be valuable for
other parties as well.
In order to sell this data, you need to create a business model. For this, you need to understand what
data people want, and who wants to buy it. Next, you have to decide on a price. Since data is already
there, the costs will be low. Lastly think of how you will send data to your buyers. Nowadays, data
marketplaces are emerging.
There are restrictions on selling data however. Personal data often should be anonymized. But even
if this is done, people could ‘guess’ who the data belonged to. There are also legal/ethical issues. You
need user permission for personal data sharing. You need to know which questions should be asked.
1.2: Selling Data: Technology issues
Personal data can contain personal sensitive information, which people do not want to be shared.
Simply removing info such as name, address, is not enough. Since quasi-identifiers combined can still
allow people to be identified. Therefore, there are anonymization measures in place:
• Data landscape analysis here you become aware of the de-anonimisation risk in your data
set and consider how an adversary would deanonymize individuals. You need to answer the
following questions:
o What info in your data set can be obtained by outsiders
o How much effort does it take to obtain the info
o How severe are the consequences of deanonymization
• Threat analysis consider the likelihood of deanonimisation. Various tools for this exist.
• Anonymization measures these are used to distort your data set. While they protect privacy,
they also distort its utility, so you need to decide how distorted the data set should be. You
can reach how much to distort based on 3 factors:
o The effort (cost, time and know-how) to deanonymize
o The likelihood of a deanonymization
o The severity of the consequences of a deanonymization in your data set
,1.3: Selling data: Ethical and Regulatory Issues
The first and most important regulation is the General Data Protection Regulation (GDPR). These
rules apply in almost the entire EU. The GDPR with personal data only.
Personal data: all data linked, directly or indirectly, to an individual called a data subject. Examples
are your address, a like on facebook and your IP.
If data cannot be qualified as personal, you do not have to follow the GDPR rules.
Processing of personal data: all activities that involve the use of personal data, includes collection,
consultation and anonymization. If anonymized correctly, the GDPR rules do not have to be followed.
In all data activities there are 2 entities, the data owner and the data controller: who processes the
data. The data controller might elocate some tasks to the processor, who has fewer responsibilities.
If there is more than one who owns the data, they are called joint controllers. To clarify who is
responsible for what, they need to sign a contract.
Personal data are not a commodity. The data subjects always have fundamental rights related to
them, even if they give it away for using a service.
There are also essential ethical principles, used to interpret the law. The European Data Protection
Supervising authority (EDPS) has developed ethical guidelines for this. One important distinction
here is the right of data subjects to not be subject to a decision based solely on automated
processing.
Buying data: Business Model Issues
A data driven business model (DDBMI) has the following characteristics:
• Data is used as a key resource
• Data analytics methods are used
• Data is part of the value proposition
How can data help your business model? 5 fundamental questions need to be answered:
• What do we want to achieve by using data?
o Use data for strategic decision making, improving processes
o Enrich core products and services with information
o Sell information offerings
• What is our desired offering?
• What data do we require and how are we going to acquire it?
o What data do we already get from existing processes? What do we get from our
products and customers and what from data providers/market places.
• In what ways are we going to process and apply this data?
o Sometimes requires expert knowledge
• How are we going to monetize it?
o Direct monetization use a subscription model, consumption based (pay-per-use) and
value-based model
o Indirect monetization, increase price of core product/service, increase sales volume
of core product or service and increase quality or reduce costs
,Buying data: Technology issues
Multi-party Computation (MPC) allows business owners to securely compute the sum of their data,
without revealing their separate data.
Issues with MPC however are that even though there might be an economic benefit, sharing the data
might be too much of a risk, due to violation of the GDPR or revealing inside information of your
business.
The aim of MPC is to enable a group of participants to jointly compute an agreed upon function,
without revealing the private input to the other participants. There is no need for a trusted 3rd party.
The most popular concept underlying MPC is secret sharing. Secret sharing can be divided into 3
steps:
• First secret shares are created
• Shares are exchanged
• The computation is done, the sums of the secrets are added
Week 1b
Why do managers need research?
(business) research is an organized, systematic, data-driven, critical, objective, scientific inquiry or
investigation into a specific problem.
Basic/fundamental research Applied
Generate body of knowledge on class of Solve specific business problems faced by
problems someone in work setting, demanding timely
solution
New contribution to scientific research Apply existing theory to solve a problem
What makes research scientific?
• Scientific statements can be verified (empirically testable!)
• Purposiveness (aim, focus)
• Testability and replicability
• Precision and confidence
• Objectivity
• Generalizability
• Parsimony (you want things found as simple as possible, but not simpler than that)
The empirical cycle
The principle of verification = you should find observations in support of your claim and start
reasoning from this empirical finding (i.e. to find ‘truth’) through induction.
The principle of falsification = a claim remains ‘true’ until proven otherwise through deduction (karl
popper).
Empirical cycle: Observation > induction > deduction > testing > evaluation >
, The research process in detail:
• Identify (broad) research problem
• Literature review
• Research objectives, questions, hypotheses
• Select research approach/research design
• Create plan for research
o Sample, instruments, data collection approach
• Collect and analyze data
• Interpret data and evaluate validity
• Communicate findings/write report
Research objective question part 1
Defining the research objective:
• Acquiring new knowledge
o To contribute to solving a problem
o To increase existing knowledge on a specific area
• The goal of research, when it is achieved
Defining the research question:
• The goal in research
• What should you know to realize the research objective?
o Should be logically deduced from the research objective
• More specific than the research objective
o How frequent is…; what is the relation between..
• Should be aimed at knowledge, not strategy/policy
• Should be empirical (not normative)
