OxCSML seminar this week: Rebecca Lewis
23 views
Skip to first unread message
Hai Dang Dau
Jan 15, 2024, 10:40:34 AM1/15/24
to oxcsml...@googlegroups.com, oxc...@googlegroups.com
Dear all,
Our first OxCSML seminar speaker of the term is Rebecca Lewis who joined the department as Florence Nightingale Fellow in September. Please find the details below. Looking forward to seeing you there.
==========
Speaker: Rebecca Lewis, Florence Nightingale Fellow, University of Oxford
Time and date: 2pm - 3pm, Friday 19 Jan 2024
Place: Room LG.03 (Small Lecture Theatre), Department of Statistics
Title: High-dimensional logistic regression with separated data
Abstract: In a logistic regression model with separated data, the log-likelihood function asymptotes and the maximum likelihood estimator does not exist. We show that an exact analysis for each regression coefficient always produces half-infinite confidence sets for some parameters when the data are separable. Such conclusions are not vacuous, but an honest portrayal of the limitations of the data. Finite confidence sets are only achievable when additional, perhaps implicit, assumptions are made. In a high-dimensional regime, we consider the implications of enforcing a natural constraint on the vector of logistic-transformed probabilities. We derive a consistent estimator of the unknown logistic regression parameter that exists even when the data are separable.
Our first OxCSML seminar speaker of the term is Rebecca Lewis who joined the department as Florence Nightingale Fellow in September. Please find the details below. Looking forward to seeing you there.
Saif & Hai-Dang.
==========
Speaker: Rebecca Lewis, Florence Nightingale Fellow, University of Oxford
Time and date: 2pm - 3pm, Friday 19 Jan 2024
Place: Room LG.03 (Small Lecture Theatre), Department of Statistics
Title: High-dimensional logistic regression with separated data
Abstract: In a logistic regression model with separated data, the log-likelihood function asymptotes and the maximum likelihood estimator does not exist. We show that an exact analysis for each regression coefficient always produces half-infinite confidence sets for some parameters when the data are separable. Such conclusions are not vacuous, but an honest portrayal of the limitations of the data. Finite confidence sets are only achievable when additional, perhaps implicit, assumptions are made. In a high-dimensional regime, we consider the implications of enforcing a natural constraint on the vector of logistic-transformed probabilities. We derive a consistent estimator of the unknown logistic regression parameter that exists even when the data are separable.
Seminar website: https://github.com/oxcsml/ML_bazaar/wiki/Seminar
===========
Reply all
Reply to author
Forward
0 new messages