Chimney fire prediction based on environmental variables

Changqing Lu*, Marie-Colette van Lieshout, Maurits de Graaf, Paul Visscher

*Corresponding author for this work

Research output: Contribution to conferencePaperAcademic

Abstract

Fire prediction and prevention is closely related to public security. In this paper, we propose a chimney fire prediction model using data collected by the Twente Fire Brigade. Random forests are used to select explanatory variables from a large set of candidates. Furthermore, we build a nested Poisson generalized linear model based on house types and weather variables. Plausible predictions are obtained, that capture the salient spatial and temporal patterns seen in the data.
Original languageEnglish
Publication statusAccepted/In press - 2021
Event63rd World Statistics Congress, ISI 2021 - Virtual
Duration: 11 Jul 202116 Jul 2021

Conference

Conference63rd World Statistics Congress, ISI 2021
Period11/07/2116/07/21

Keywords

  • areal unit data
  • conditional variable importance
  • Poisson generalized linear model

Fingerprint

Dive into the research topics of 'Chimney fire prediction based on environmental variables'. Together they form a unique fingerprint.

Cite this