Project

Name:

Adjustment for existing application(s)

 Yes, name:

 No

Client details

Organizational unit client

Cost center number client

Client (budget manager)

Client role

Delegated client

Role delegated client

Contact person client

Role contact person

Relationship with other project(s)

 Yes, name:

 No

Project description

Briefly describe what the client wants to see realized.

Description of business unit

Briefly describe the function of the business unit for which the application is intended.

Non-functional requirements

Number of expected users

<10 | 10-50 | 50-100 | 100-250 | >250

Expected frequency of use

ad-hoc | daily | weekly | monthly | quarterly | yearly

expected mode of use

Standard reporting | Search interactively

Up-to-datedness of the data (refresh rate)

ad-hoc | daily | weekly | monthly | quarterly | yearly

Addition and or explanation

Data security check

A data security incident is a serious security breach. When protected, sensitive, or confidential data is stolen, used, viewed, stored, or transmitted by someone who is not authorized, a breach occurs. Other security concerns include data leaks, information leaks (who can see what), sensitive information leaks, and data spillage. As a result, when it comes to data and information usage, we must identify some specific elements.

Application/model/app

Dashboards and reports are critical for any organization, team, or department. The work cannot be completed without this information.

Essential | Important | Desirable | Unclassified

Confidentiality

The dashboards and reports are private and only available to a select group of managers and employees.

Essential | Important | Desirable | Unclassified

Access Control

What type of access control is required when confidentiality is required? What are the rules that we must follow? (Manager of team A – can see information about team A, and so on).

Add information:

Integrity

Incorrect information will not be tolerated (accountable environment). The organization suffers significant damage if the information is incorrect, incomplete, or late. Inaccurate data, such as inaccurate financial transactions, undermines trust in the organization.

Essential | Important | Desirable | Unclassified

Availability

Dashboards and reports are essential. The organization will suffer severe consequences if the information is not available.

Essential | Important | Desirable | Unclassified

Addition and or explanation

Data privacy check

Some countries have special legislation and regulations when it comes to data usage and data privacy. For example, in the Netherlands, there is a regulation called General Data Protection Regulations (GDPR). When it comes to regulations in the US it is a patchwork of rules. There is actually no specific privacy law except in relation to children but in the State of California, they have implemented data privacy regulations.

When it comes to using personal data we should take care and be aware upfront that we need to consider the privacy regulations for our project. There are several categories that we have to identify and classify.

Name, address, and city data

(Name, address, postal code, city of residence)

 Yes

 No

Identification data

(Passport, driver’s license, or social security number)

 Yes

 No

Application details

(Application letters, resume)

 Yes

 No

Contact details

(E-mail address, phone or fax number)

 Yes

 No

Salary details

(Salary details, social payments, income taxes, expense reimbursement)

 Yes

 No

Social media accounts

(LinkedIn, Twitter, WhatsApp, Facebook)

 Yes

 No

Image and sound recordings

(Video, photos, passport photographs)

 Yes

 No

Payment details

(Bank name, account number, name of account holder)

 Yes

 No

Addition and or explanation

When a checkbox is answered with “yes” it is advised to arrange a legislation and regulations check with your Privacy Officer.

Legislation – regulations check

Is privacy impact analysis needed?

Essential | Important | Desirable | Unclassified

Is a register of data needed?

Essential | Important | Desirable | Unclassified

Addition and/or explanation

Data ethics checklist

Data generation, use, collection, analysis, and dissemination are all aspects of data. We can do this with both unstructured and structured data. When we do so, there is a chance that the decisions we make will have an impact on individuals and the world. As a result, we must be open about how we use data in our projects.

1

Do we have to take care of laws and regulations in this data and analytics project?

The first critical step is to determine whether any legalizations or regulations are applicable to the project.

 Yes

 No

Write any additional information here:

2

Is the data that we want to use available in an ethical manner and is this data suitable for usage?

We must be mindful of who owns the data and ensure that it is used in the manner intended by the owner!

 Yes

 No

Write any additional information here:

3

Is it possible to identify and check bias in the data that we have collected or used for our models?

People can be biased by their origin, and the same is true for data and the application of algorithms and business rules. The data we collect and store is not as objective as we believe! When using algorithms and business rules, we should be aware that the data we use to train the models can have an impact on people and possible human bias can be magnified, which results in undesirable outcomes.

To conclude, we must be able to identify, test, verify, and discuss the results.

 Yes

 No

Write any additional information here:

4

Can we identify and demonstrate bias in our created model or in the data used?

When we use data and apply various models, we may have used data that is biased. When we use that specific data set and apply the learning models, the model produces biased results. For example, a bias based on gender, age, equality, or racial elements.

We need to be aware that we must consider documenting, discussing, and evaluating our data usage choices. The message here is to avoid doing things simply because you can!

 Yes

 No

Write any additional information here:

5

Can the legal rights of individuals be impinged by the use of data?

When an individual’s legal rights are at stake, the organization must have permission and the right to use data for specific purposes. As an example, suppose an organization provides data to its direct partners, but different privacy conditions than the internal data usage may be addressed here. In the event of an incident, for example, they know the addresses and more detailed information from people, but certain data is not shared to protect those people’s privacy (such as names, address details, and other things on which someone can be individually identified).

It could also happen internally, for example, by logging information that is known at the employee level but is shared with users at the department or concern level. We should be aware of this in order to protect the privacy of each individual employee.

 Yes

 No

Write any additional information here:

6

Are we able to understand that the data we want to use is suitable for the purposes of our project?

When we begin an analytics or data science project, we must understand and ensure that the data we intend to use is appropriate for the purpose of our projects. Following that, we should be able to verify and validate the data for our project.

For example, when records or values are missing, the outcome or our algorithms and business rules can have a significant impact on the results, potentially producing a biased result.

 Yes

 No

Write any additional information here:

7

Do we have a multi-disciplinary team present to discuss the present dilemmas, and explore the possible usage of algorithms and the possible outcomes?

When it comes to assessing and discussing our own work, we need to focus on the dilemma and the outcomes that can occur with a multi-disciplinary team.

 Yes

 No

Write any additional information here:

8

Explainable AI by design? Are we able to define the role of an algorithm used and what processes are being followed (procedural transparency)?

It is critical for data engineers and data scientists who train models to understand the model’s behavior in order to detect errors or weaknesses. This is why we must correctly describe the used algorithms or business rules.

When data scientists and data engineers train a model, it is critical to understand the model’s behavior. They must be able to identify any flaws or errors.

 Yes

 No

Write any additional information here:

9

Explainable AI. Are we able to explain the algorithm or business rules to the guardians, stakeholders, and others whom it concerns?

Explainable AI is defined as the ability of a person to comprehend the reason for a decision. The decision is influenced by algorithms and business rules. To be able to understand the model’s decision, we must be able to explain the decision. We can do so by design, but we can also do it post-hoc by using an algorithm to understand the black-box model.

 Yes

 No

Write any additional information here:

Addition and/or explanation

Are organizational changes to be expected that could affect the progress of the project?

 Yes

 No

Has a project manager/leader/coordinator been appointed on the client side?

 Yes

 No

Is the contact person present full-time? Or is there a full-time backup?

 Yes

 No

Can a response time to questions of a maximum of 3 days be guaranteed?

 Yes

 No

Does the contact person have knowledge of the source systems and the source data?

 Yes

 No

Is all data available in the source systems?

 Yes

 No

 No, but the development/adaptation of the source systems is planned and will be released on the following date:

Are there interfaces with the source systems?

 Yes

 No

 No, we have to arrange this

Are there dependencies with external suppliers?

 Yes

 No

Are there dependencies with external suppliers?

 Yes

 No

Source system data (per source system)

Source system name

Source system type

 Database

 Interface, an automated connection

 File, manual (e.g. MS Excel, CSV, and so on)

 Cloud

 Other

Owner source system (department)

Internal/external management organization

 Internal – name:

 External – name:

Contact person

Is the source system data model available?

 Yes

 No

Are field definitions available?

 Yes

 No

Particularities

Size of data delivery (number of files)

Number of expected rows

Additional information

Source system data (per source system)

Source system name

Source system type

 Database

 Interface, an automated connection

 File, manual (e.g. MS Excel, CSV, and so on)

 Cloud

 Other

Owner source system (department)

Internal/external management organization

 Internal – name:

 External – name:

Contact person

Is the source system data model available?

 Yes

 No

Are field definitions available?

 Yes

 No

Particularities

Size of data delivery (number of files)

Number of expected rows

Additional information

Other project data

Project number

Date of intake

The appointed project manager or project leader

Date intake forwarded to the contact person

Name of the person who did the intake

Relationship with other projects?

 Yes, name of project:

 No

Describe the circumstances under which the your project took place (reason, parties involved, where, etc.)

Describe the exact assignment you worked on (and who was your project leader, client ), or that you assigned to yourself. Indicate what your role/position was in this project

Describe the approach you used (activities; tools; preparation)

How was the result assessed, by whom and on what basis?

Why do you consider this your best practice, how do you look back to the work that you have done? Relate to the complexity of the situation and/or development in feedback you’ve received.

Name

Option 1

Description

Pros

Cons

Strategic contribution

Name of KPI

Definition of KPI

Owner of KPI

What is the purpose of usage?

Frequency

Department

Type of graph

Additional information

Unit of measure

Data source

Required tables

Reporting period

Data owner

Data steward

Date of KPI approval

Norm

KPI