Bio
Azam Ali is a PhD Student at the Institute for Transport Studies, University of Leeds supervised by Charisma Choudhury, Stephane Hess, and Ed Manley. His PhD research is related to addressing the challenges in utilising passively collected sources such as smartphone-based travel surveys to develop travel demand models. His research interests lie in the use of discrete choice models and machine learning techniques to understand and model travel behaviour. Prior to starting his PhD, Azam carried out a Masters in Transport Planning and Engineering and Masters in Social Research at the University of Leeds. He has previously compared and contrasted discrete choice models and machine learning techniques to model vehicle ownership decisions and time-use.
Research interests
Azam’s PhD is related to addressing the challenges in utilising passive data sources such as call detail records and smartphone-based travel surveys to model travel behaviour. Smartphone-based travel surveys rely on algorithms that infer trip details such as trip modes and purposes from GPS traces, which allow researchers to collect large amounts of data with minimal input from the respondents. To ensure that the on-ground truth is collected, there is a need for users to validate the inferred modes and purposes. However, a lot of users fail to validate leading to large amounts of unvalidated data. In his first study, Azam develops a modelling framework which models daily trips using unvalidated trip purpose data. The research finds that it is better to impute probabilistically rather than deterministically, and it is feasible to generate reliable travel demand models without the need for validation. In his second study, he uses posterior analysis, which is a Bayesian approach, to better predict choices for individuals who have previously tagged their trips. The proposed approach outperforms other imputation techniques. At Turing, he plans to work with passive data sources to model and understand travel behaviour. Further, he aims to learn more about new methodologies especially machine learning approaches to better model travel behaviour.