Antony Overstall - Approximate Bayesian Multiple systems estimation with partial classification (- in person)
| Dates: | 26 April 2023 |
|---|
| Times: | 14:00 - 15:00 |
|---|
| What is it: | Seminar |
|---|
| Organiser: | Department of Mathematics |
|---|
| Who is it for: | University staff, External researchers, Current University students |
|---|
| Speaker: | Antony Overstall |
|---|
|
Antony Overstall, Associate Professor in Statistics in the Department of Mathematical Sciences at the University of Southampton is our speaker for the Statistics seminar series.
Title: Approximate Bayesian Multiple systems estimation with partial classification
Abstract: Multiple systems estimation (MSE) refers to statistical methodology used to estimate unknown and elusive human population sizes, from administrative data. These population sizes usually consist of vulnerable people, e.g. individuals subjected to human trafficking or who are injecting drug users. Typically, in MSE, individuals from the target population are observed on a series of administrative lists (e.g. police records, hospital records, etc). Individuals can appear on more than one list and so these lists are matched so that the count of individuals observed on each combination of lists is determined. These data can then be used to estimate the unknown total population size and quantify uncertainty in this estimation. In some cases, it is not possible to match between certain lists, i.e. partial classifiication. Subsequently, only linear combinations of certain counts are observed, which can render the likelihood computationally expensive to evaluate. Borrowing a saddlepoint likelihood approximation from the frequentist MSE literature, this talk develops a Bayesian MSE approach for partial classification which can account for model uncertainty and can be extended to other missing data MSE problems.
Venue:
Frank Adams Seminar Room 2
Alan Turing Building
Manchester
M13 9PL
Speaker
Antony Overstall
Organisation: University of Southampton
Travel and Contact Information
Find event
Frank Adams Seminar Room 2
Alan Turing Building
Upper Brook street
Manchester
