Supercrunchers: the human in an era of data science

More and more data about people and organizations are being collected and present-day computing power also allows such data to be analyzed and used for analytic and predictive purposes. Although in many cases actual decision-making is based on data, there are still surprisingly many instances where using relatively straightforward models works just as good or better than having a human expert decide.
The fields of application vary widely. That the price of a bottle of Bordeaux can be adequately predicted on the basis of just four straightforward indicators (outperforming expensive and experienced experts), is in itself a rather innocent finding. But similar cases can be made for life-and-death decisions: there is evidence that whether you are having a heart attack or not can be predicted quite accurately using simple models, just as good as experienced physicians. Still, even in cases where the superiority of the model-based prediction is undisputed, we see that it is not always implemented. The relevant questions that are treated in the course is when and how model-based decision making can be used, how and when it is and can be implemented, and what the consequences are for the involved actors and society as a whole.

The idea that data can be used, somehow, to uncover knowledge and insight has been the subject of several popular scientific books: “Supercrunchers”, “Freakonomics”, “The tipping point” and “The signal and the noise” are just a few. The course consists of two parts: one part with lectures in which we cover several examples from the (popular) literature and consider their merits in the knowledge economy in more detail. The second part consists of analyzing a case in which students will perform some actual data crunching on a real life example. Besides being of interest for HTI-students with an interest in decision-making and data, for Innovation Sciences students with a focus on aspects of the knowledge economy, but also for web or data science students with an interest in human decision-making.

The course aims to give students an in-depth understanding of the literature on model-based versus human decision making, with an emphasis on the way in which humans and models deal with large amounts of information (“big data”). In addition, students learn hands-on what the typical pitfalls of dealing with (large scale) information are. This includes both the knowledge of the general process characteristics of adequate data scientific efforts, and a general sense of the ideas and prejudices of humans involved in such efforts.

After this course, students are able to:

• Compare the principles of model-based and human decision-making, particularly in contexts involving large-scale information and big data.
• Analyze and Evaluate real-world examples from literature and case studies to determine the merits and limitations of model-based versus human decision-making approaches.
• Apply data science methodologies to perform hands-on data analysis on real-life cases, identifying common pitfalls in handling large-scale information.
• Critically assess the impact of model-based decision-making on societal and individual actors, exploring ethical, practical, and implementation challenges.
• Discuss the general process characteristics of successful data scientific efforts, including how biases and human prejudices can influence decision-making outcomes.
• Design and Conduct a case study, using data-crunching techniques to develop insights and predictions, while articulating the strengths and weaknesses of the approach.
• Compare the effectiveness of simple predictive models versus human expertise in various fields, ranging from economics to healthcare, using evidence from the literature.
• Formulate actionable recommendations for implementing model-based decision-making in specific contexts, considering potential barriers and societal implications.

Je moet voldoen aan één van de onderstaande verzamelingen met eisen

• Verzameling 1
• Bachelor of Science (BSc) afgerond
• Verzameling 2
• Schakelprogramma afgerond

• Reader containing material from several books and journal articles