Data Science and Society

Level: Master’s
Institution: Utrecht University
Teaching period: Term 1 (September–November)
Student background: Information Sciences and other interdisciplinary Master’s programmes
Project format: Teams of approximately five to six students
Core technologies: Python, pandas, scikit-learn, Plotly, databases, APIs, Docker and related deployment tools

By the end of the course, students should be able to:

• explain the role of data science and examine its potential societal impact;

• recognise knowledge-discovery processes used in applied data science;

• identify relevant developments in data science and machine learning;

• apply selected statistical and machine-learning methods to real-world problems;

• analyse and preprocess different forms of data;

• distinguish between descriptive, predictive and optimisation-oriented analytical tasks;

• retrieve and integrate data from files, databases and APIs;

• use simple database queries within a practical data-science system;

• apply containerisation techniques within a team project;

• explain principles of privacy-preserving data science and responsible AI;

• consider ethics, privacy and explainability when designing data products;

• apply the phases of CRISP-DM throughout a complete data-science project.

RESOURCES

INFOMDSS Project Ideas

Explore project ideas for real-time data science dashboards addressing meaningful societal or organisational problems. Each idea helps you consider the intended user, possible KPIs, suitable data sources, and realistic project scope.

KPI Worksheet

Define the user, desired outcome, and driver and lagging KPIs before selecting datasets or designing charts. Use this worksheet to identify weaknesses in your project idea while they are still easy to address.

DSS Starter Kit

Start your project with practical planning resources, technical guidance, and reusable templates. The starter kit helps your team move from an initial problem to a feasible, well-structured dashboard project.

APPLIED DATA SCIENCE PROJECT

The learning objectives of this project target primarily the levels of application, analysis, and evaluation in the Bloom's framework. The creation of new methodology is not within the scope of this course. These levels are applied on all stages of the CRISP-DM process, from business understanding to deployment. Therefore, at the end of the project, students should have learned how to

• plan a data science project on a strategic level, e.g., identifying a use case and the key users and requirements therein, as well as relevant data sources, (performance) indicators, etc.;

• organise data, for example in a relational data base;

• import data from various sources, e.g., reading from CSV files, querying from a DBMS, fetching from a webservice via SOAP (if/as needed);

• join and manipulate data with pandas;

• perform descriptive and predictive analytics;

• visualise data, for example with Matplotlib;

• deploy the project, including reporting;

The practical team project complements the theoretical parts of this course. The practical team project accounts for 40% of this course. This corresponds to 90 individual and overall 540 person working hours within each 6-personteam (approx. 10 person weeks).

Past Project Examples

Monitoring residential solar-panel performance

One project developed a system for monitoring domestic solar-panel installations and identifying indications that a system was underperforming or experiencing a technical problem. The challenge required the team to distinguish normal variation from patterns that could justify inspection or maintenance. The dashboard was designed to help users understand performance over time and diagnose potential problems.

Planning air travel during changing COVID-19 restrictions

This project supported travellers planning international journeys during rapidly changing pandemic restrictions. The system combined information about travel restrictions with predictions of potential increases in COVID-19 cases at particular locations. Users could compare possible destinations and timing while considering both current restrictions and anticipated changes.

The project illustrated the difficulty of building decision-support tools from uncertain, time-sensitive and changing data.

Evaluating environmental initiatives across locations

Another project developed a dashboard for examining where environmental initiatives appeared to be producing results and where progress remained limited.

The system allowed users to compare locations, indicators and interventions. The project required careful definition of success measures because environmental outcomes may depend on multiple contextual factors and cannot always be attributed to a single intervention.