SURF Research Poster

Overview

An undergraduate research poster produced for the SURF (undergraduate research) program, mentored by a faculty mentor at ASU's College of Liberal Arts and Sciences. The project analyzes Residential Energy Consumption Survey (RECS) data in R to uncover relationships between household variables and energy burden in Arizona. It frames the research question, methods, findings, and policy implications around how income and assistance status relate to energy burden.

Key Achievements

• Framed and investigated the research question of how RECS data can demonstrate connections between household variables
• Applied data mining and regression analysis in R to the RECS dataset
• Found that energy burden in Arizona skews toward lower values, with the median falling in the lower quartile
• Identified that lower-income households and those receiving assistance show higher energy burden percentages
• Documented analysis obstacles and mitigations (missing-data imputation via k-NN/MICE, dimensionality reduction via PCA, nonlinear modeling via Random Forests/Neural Networks)
• Produced supporting visualizations including an energy burden distribution histogram with density overlay and a combined income-vs-burden scatter graphic

Approach

The work uses R for data mining and regression on RECS data to surface relationships among household variables. Energy burden is examined through distributional analysis (a histogram with a density-curve overlay, outliers removed) and a scatter plot of energy burden against household income and assistance status. The poster also outlines planned techniques for handling missing data, high dimensionality, and nonlinear relationships.

Tools & Technologies

Gallery

Results

Analysis indicates Arizona's energy burden disproportionately affects lower-income households, particularly those not receiving assistance, motivating targeted policy intervention; see docs/SURF POSTER.pptx for the full poster and figures.