Nonlinear Time Series Analysis and Manifold Learning Laboratory
Give students the experience of an end to end experience in modern data driven science from problem approach to paper submission
The goal of this course is to select a group project and take it to completion among the students of the class.
The course will encompass a complete end to end project from Data selection and posing the scientific questions to submission of a complete manuscript to a peer reviewed journal as well as to a preprint server.
The purpose of the project is to teach the students how to apply the techniques that they learned in the previous term in a real life analysis problem. The scope of the class is broader as it also aims to impart instruction on how to choose a scientific problem and the data that will allow answering such question. In addition students will learn best practices for scientific narratives in addition to data driven problem solving .
Students who have not taken A111 will most likely not be prepared to take this class because the relevant material is rarely offered in all but maybe 2 or 3 places in the world.
1. Data selection
2. Scientific question formulation
3. Data suitability assessment
4. Data properties
5. Causal inference and network structure
6. Evidence accumulation
7. Scientific outcomes
8. Manuscript crafting
9. Submission process
10. Scientific integrity and standards of rigor
100% class participation and contribution to the group paper.
Required pass in first theoretical portion of this course, A111 Nonlinear Time series Analysis and Manifold Learning.
Prior deep knowledge of Taken’s theorem-based methods is an absolute prerequisite.
Analysis of Observed Chaotic Data, Henry DI Abarbanel; Springer
Data-Driven Science and Engineering: Machine Learning, Dynamical Systems, and Control 2nd Edition, by Steven L. Brunton and J. Nathan Kutz
Alternate years: AY2026
Follow-on course from A111 (required)