The Behavioral Data Science lab is a research group in the UTS Data Science Institute at the University of Technology Sydney, Australia.
Our research aims to make sense of large amounts of behavioral data, to understand online behavior and its offline effects. We develop core methods in machine learning and optimization in order to distill heaps of raw data into meaningful insights in areas including
If you're interested in working with us, get in touch!
Summary: We build a machine learning-based Job Transitions Recommender System that can accurately predict the probability of transitioning between occupations. We showcase the system for workers forced to transition between jobs. The system is based on a novel data-driven method to measure the similarity between occupations based on their underlying skill profiles and real-time job ads. We also build a leading indicator of Artificial Intelligence adoption in Australian industries, outlining gaps, opportunities, and trends.
We demonstrate in this blog post a tutorial on applying the tools for analyzing online information diffusions about Twitter users, birdspotter and evently. Dataset In this tutorial, we apply two tools for analyzing Twitter users, on a COVID-19 retweet dataset. The dataset is curated by Chen, et al. One can obtain a copy of the tweet IDs from their project. We only use the 31st of Janury sample of the whole dataset for demonstration purpose.
Most statistics students will be familiar with the phrase “correlation isn’t causation,” however, this doesn’t feature strongly in the remainder of their educations. To overcome this hurdle, the researchers’ best practice in experimental design is the randomized controlled trial. However, there are only specific experiments that we can perform. For example, to test the whether smoking causes cancer, we can’t force subjects to smoke. ⊕In the 1950s the tobacco companies argued that there could be some confounding factor (a gene) which smokers and lung cancer patients shared.