Statistical Machine Learning Projects

Predicting housing prices in King County, Washington

In this project, I built statistical machine learning models to predict housing prices from a data set of houses located in King County, WA. The goal was to predict house prices based on features of the houses, including square footage, the number of bedrooms and bathrooms, and condition, among others. I also created graphical visualizations of the data and plotted relationships among the variables using ggplot in R. You can view a short report of my work here or view the entire project in my github repository.


Academic Research

Context-based person impressions

In one line of research (Huang, Sacchi, & Sherman, 2017), we studied how people form impressions of another person when that person is encountered in different contexts. Drawing from the attention theory of category learning (Kruschke, 2001), we have shown that impressions of a target in a rarely occurring context are learned after impressions in a frequently occurring context, and that this learning order leads to stronger rare context impressions than common context impressions. This process occurs when people form both trait impressions and evaluative impressions.

Expectancy maintenance in context-based impression formation

Oftentimes, people’s attempts to maintain consistency in their impressions of others result in impression formation that is biased toward expectancy maintenance. In my dissertation research, I studied how people use context as a means of maintaining expectations of an individual target’s personal characteristics when that target is observed across multiple contexts. We showed that people generalize expectancy-confirming impressions across contexts whereas they contextualize expectancy-disconfirming impressions to the unique context in which the target’s expectancy-disconfirming behaviors are observed. By contextualizing expectancy-disconfirming behaviors, people limit the ability of those behaviors in changing their impressions of the target outside of that single context. This is true whether the expectancies are based on individual characteristics or on group stereotypes. This research has implications for stereotyping and attitude change. An outline of this research and the R code for data analyses are available in my github repository.

Perceptions of college students with disabilities

In this project, we examined how college students perceive their peers who have a disability and whether these perceptions differ depending on the visibility of the disability. We showed that students with visible disabilities (i.e., physical disabilities) compared to students with non-visible disabilities (i.e., cognitive and psychiatric disabilities) were perceived as being more sociable and academically capable, but that all were equally deserving of academic accommodations. Understanding the perceptions of students with disabilities is one step toward creating a more inclusive and welcoming campus environment.