Masters candidate in Biostatistics, Karl Brown, will present:
“Comparing Clustering Algorithms to Categorize Groups of Individuals with Autism Spectrum Disorder in Vocational Rehabilitation Programs”
Plan B Adviser: Erika Helgeson
Abstract: Employment is a facet of life that allows one to provide for themselves financially, but it also gives one a sense of purpose and meaning. Vocational Rehabilitation (VR) programs across the United States assist those with disabilities and impairments to obtain employment opportunities. An exponential increase in the number of children being diagnosed with Autism Spectrum Disorder (ASD) has emphasized the need to characterize the population of individuals with ASD using VR programs. This analysis uses data collected from VR programs, with a focus on those with ASD, to better understand characteristics of people who enter VR programs and what makes an individual more likely to exit a program with a job. This is accomplished by employing 4 machine-learning clustering algorithms to group individuals and a series of multiple logistic regression models which assess how the likelihood of getting a job is affected by individual level characteristics. Almost all characteristics explored had a significant effect on the likelihood of getting a job. Each clustering method produced clusters that were distinct from other methods, but each formed clusters that were based on similar characteristics such as age, education level, and funding source. Ultimately this analysis reveals that basic characteristics such as age and education level can be very telling about the success an individual may find in a VR program and categorical clustering methods are perhaps best utilized when they are used in tandem as they each reveal different perspectives of the groups that exist in a dataset.