Data analytics is at the heart of data science and AI applications that are rapidly reshaping our society and its.
The Data Analytics Unit (DAU) of JADS combines expertise in the domains of causal inference, social network analysis, AI, and in particular machine learning with the goal of explaining the dynamics of complex phenomena. The members of DAU study, develop, apply and evaluate data analysis methods “for a wide variety of data-intensive application domains.” door “that allow individuals, organizations and society at large to make better decisions and pursue beneficial goals. for a wide variety of data-intensive application domains. Our conception of a data analysis method is any formal, statistical, or machine learning method that can be trained on real data for making predictions in an application domain and allow researchers to identify leverage points that are crucial for understanding systems behavior.
Our expertise in causal (probabilistic) inference contributes to furthering our understanding of the strengths and limitations of the data analysis methods in data science and AI. For instance, the data analysis method at the core of the “AI revolution”, deep learning, has an excellent prediction performance but lacks a proper understanding of why it works. Probabilistic methods can complement the predominant empirical approaches to deep learning, thereby improving the transparency and reliability of data-science and AI applications.
Social network analysis contributes to the integration of data science and AI technology in our society. Social Systems are progressively shifting to large scale employment of data science and AI. Understanding the way in which social communities take decisions (policy makers, health organizations, and industry) by interacting with each other and with technology is pivotal to this transition. Data science tools are complementary to study and analyze social networks.
In the area of recommender systems, we go beyond accuracy metrics, evaluating these systems also on their behavioral and experiential aspects. How do algorithmic aspects (e.g. an objective metric of diversity or novelty) relate to perceptions of diversity and novelty? do recommendations provide more satisfying and better experiences, do they help people to change their behavior: for example in food recommendations, how can we help them find more healthy food that fits their preferences, or in music: how can we help them to explore new genres or change their mood in a personalized way?
Joran Lokkerbol is a data scientist and director of Centre of Economic Evaluation & Machine learning at Trimbos. Furthermore has he specialised in machine learning at Harvard and MIT, and is active as a post-doc at the UMCG. Issues that interest him are optimization of the healthcare system, consultancy and the development of models. In recent years his main focus has been evaluating cost effectiveness of interventions in mental healthcare.
Our ambition is to support our researchers in setting up or expanding their research groups by acquiring grants that fit their expertise and ambitions. With regards to application domains, we have the advantage that data analytics methods can be applied to virtually every domain. Given the three application domains spearheaded by JADS, we will actively seek to contribute to Crime and Safety, AgroFood and Biodiversity, and Social Entrepreneurship.
In addition, we would like to continue and possibly expand our successful collaboration with CZ. We also intend to collaborate with Wageningen university, locally the HAS, and partners from industry.
Data engineering for studying sensor networks and machine learning pipelines.
Using machine learning and network analysis to study interactions in meetings, entrepreneurial pitches, social influence, interaction norms and routines.