About the event
Experimental design choices influence both what a scientist can discover, as well the confidence that can be placed in the final outcome. Poor experimental design decisions can also lead to wasted time and resources, as well as non-unreproducible research.
The emerging area of data science for experimental design aims to develop computational strategies to design experiments in a principled fashion, exploiting data science to produce more efficient, more accurate, and more reproducible research. In this workshop, we will explore how data science can help us design, perform, and analyse scientific experiments. The workshop will cover aspects of experimental parameter optimisation, laboratory automation, and issues around reproducibility of data analysis. It will include a discussion session about the barriers to incorporating data science for experimental design in real laboratories, led by social scientists who study the behaviour of experimental scientists.
The aim will be to bring together data scientists who work on data science for experimental design, experimental scientists/practitioners, and social scientists to examine the latest research in the field, as well as to discuss practical steps towards integrating these tools in the laboratory.
Morning session 1: Planning experiments: How can we use data science to design experiments to be as informative as possible?
Morning session 2: Conducting experiments: How can we use laboratory automation and robotics to reduce research biases?
Afternoon session 1: Analysing experiments: How can we make research more reproducible?
Afternoon session 2: From theory to practice: What are the obstacles to integrating data-driven experimental design in real laboratories, from an ethnographic and sociological perspective?