Data Science Special Lecture -2 (2022)

Date & Time: Wednesday, 12 October 13:30-15:00
Place: L2（IS）, L3（IS) and online(webex)
Language: English

Lecturer: Prof. Ichiro TAKEUCHI (Graduate School of Engineering Mechanical Systems Engineering, Nagoya University)

Title: Statistical Reliability Quantification for Data-Driven Scientific Discovery

Abstract:
Data-driven science is a new scientific paradigm that aims to use data and algorithms to gain scientific knowledge and insights, with the potential to uncover new scientific discoveries that cannot be easily recalled from expert knowledge and experience. On the other hand, data-driven hypotheses are generated by transforming complex data with complex algorithms, making it difficult to evaluate the reliability of the hypotheses within the framework of traditional statistical inference. In this lecture, I will explain how the statistical reliability of scientific findings should be evaluated in data-driven science, and will introduce a method called selective inference, which has been recently developed for evaluating data-driven hypotheses.