Accelerate Science’s ‘Data Pipelines for Science’ School

How can researchers design and implement data pipelines for scientific research? Our 'Data Pipelines for Science' School will help scientists to learn how to correctly, efficiently and robustly prepare your datasets for machine learning in your scientific projects.

Well-curated and managed data is central to the effective use of AI, in science and elsewhere. How can scientists build the data pipelines they need to accelerate their research with AI?

The Data Pipelines for Science school was originally launched in Winter 2022 and three cohorts have now participated in the course.

Machine learning is an important  tool for researchers across disciplines. Scientists today have access to more data, from a greater range of sources and at greater speed than ever before, and opportunities to extract insights from this data using AI. But before deploying AI, researchers must have a data pipeline that transforms their data into a state that is suitable for the machine learning algorithms being used. 

These pipelines are important independent research outputs as they enable others to easily inspect, reproduce, refine or extend the scientist’s work. However, implementing data pipelines present numerous software challenges that might be difficult to resolve or even identify to scientists who do not have a significant expertise in software engineering concepts and practices. 

Such challenges include: how do I ensure the correctness of my pipeline? How do I structure my pipeline in a way that makes it easier for others to reuse and extend? How do I ensure my pipeline is robust enough to deal with different types and volumes of data? How do I document and publish my pipeline? How do I ensure my pipeline adheres to privacy and anonymisation constraints?

Accelerate Science’s ‘Data Pipelines for Science’ School helps scientists overcome such data pipeline challenges by equipping them with the latest best-practice software techniques. It consists of a blend of lectures and labs, with a focus on discussing general principles and case-studies during the lectures, and a focus on hands-on exercises in Python during the labs. Participants also have the opportunity to discuss and share data pipeline issues encountered in their own research with the course instructor and cohort, and to relate it to the course content.

Data Pipelines School materials

For a taster of topics covered in you can find materials from previous Data Pipelines School previous Schools, check out the Accelerate Github account. You can also watch watch Dr Soumya Banerjee’s talk on Reproducible Research in R here.