Loading Events

« All Events

  • This event has passed.

SMART Seminar Series

December 15, 2017 @ 2:00 pm - 3:00 pm

Free

This seminar has now passed

View seminar photos on flickr View seminar slides on SlideShare

Presenter – Morgane Dumont

MORGANE DUMONT BIO

Morgane Dumont is a PhD student in applied mathematics. She is interested in statistics, microsimulation, agent-based modelling, machine learning and synthetic populations. For her research, she developed a static synthetic population for Belgium in 2011 as input for an agent-based modelling aiming to forecast the health needs of the elderly for Belgium in 2030.

She co-authored the book “Spatial Microsimulation in R” with Dr Robin Lovelace to help new researchers to dive faster into the subject and learn how to code microsimulation in R.

She is collaborating with SMART and Dr Johan Barthélemy within the fields of agent-based modelling and machine learning.

Finally, she collaborated with Dr Alexis Petit, Dr Daniel Casanova and Prof Anne Lemaître to develop a synthetic population of space debris.

How to improve the order of evolutionary models in agent-based simulations for population dynamics

Abstract

Agent-based modelling is used more and more in various fields, such as transportation, social sciences, health cares. These models are composed of several modules that can be consider as sub-models. For example, the sub-models in TransMob (Huynh, 2015) include death, birth, ageing, marriage, divorce.

Our current research questions some of the usual choices made in the development of such models with a focus on the ordering of the sub-models. Indeed, the agent-based models detailed in the literature perform the simulation in an arbitrary order without proper justifications for the retained order.

Our main goal is to first determine if the order significantly impacts the results, and if so, identify patterns in orders resulting in similar final populations. The model TransMob developed at SMART is used to simulate the population during 10 years for each possible order of the models. We show that defining the order is crucial and propose a method allowing a reduction in the possible combinations and a decrease the variance of the simulated populations.

RSVP

Tickets are not available as this event has passed.

Details

Date:
December 15, 2017
Time:
2:00 pm - 3:00 pm
Cost:
Free
Events Category:
Events Tags:
,
Website:
http://smart.uow.edu.au

Organiser

SMART Infrastructure Facility
Phone:
+61242215031
Email:
smart-facility@uow.edu.au
Website:
http://smart.uow.edu.au/

Venue

Building 6 Room 105
SMART Infrastructure Facility | University of Wollongong
Wollongong, NSW Australia
+ Google Map