Decision Tree Model of Smoking Behaviour

Abo-Tabik, Maryam, Costen, Nick ORCID: https://orcid.org/0000-0001-9454-8840, Darby, John and Benn, Yael ORCID: https://orcid.org/0000-0001-7482-5927 (2020) Decision Tree Model of Smoking Behaviour. In: 5th IEEE Internet of People and Smart City Innovation 2019, 19 August 2019 - 23 August 2019, Leicester, UK.

Accepted Version Available under License In Copyright. Download (1MB)

Abstract

Smoking is considered the cause of many health problems. While most smokers wish to quit smoking, many relapse. In order to support an efficient and timely delivery of intervention for those wishing to quit smoking, it is important to be able to model the smoker’s behaviour. This research describes the creation of a combined Control Theory and Decision Tree Model that can learn the smoker’s daily routine and predict smoking events. The model structure combines a Control Theory model of smoking with a Bagged Decision Tree classifier to adapt to individual differences between smokers, and predict smoking actions based on internal stressors (nicotine level, with- drawal, and time since the last dose) and external stressors (e.g. location, environment, etc.). The designed model has 91.075% overall accuracy of classification rate and the error rate of forecasting the nicotine effect using the designed model is also low (MSE=0.048771, RMSE=0.216324, and NRMSE=0.153946) for regular days and (MSE=0.048804, RMSE=0.216637, and NRMSE=0.195929).