About the Human Data Science group
The Human Data Science (HDS) group, led by Dr. Daniel Oberski, operates at the intersection of statistics, computer science, and social and behavioural sciences. Data is often about humans, and data about humans is messy. As a result, data science is prone to errors at every step of the investigative cycle. With this in mind, the HDS group develops methods which:
- Acknowledge that errors – in data, modelling decisions, and implementation – are inevitable;
- Study the degree to which errors actually affect the output;
- Remove these effects where necessary, either by preventing the errors or, when this is not possible or cost-effective, by correcting for their effects.
Different fields have – sometimes independently – generated much knowledge about these topics. Therefore, the HDS group works in a variety of domains, including latent variable models, natural language processing, measurement models, missing data, high-dimensional data, and more. The HDS group is embedded in the Department of Methodology and Statistics with strong ties to Utrecht University’s focus area of Applied Data Science.