Andrea Figueroa

I'm currently a Research Assistant and PhD Student in Human Centered Design & Engineering (HCDE) and a member of the Human-Centered Data Science Lab at University of Washington, Seattle. I received my bachelor's and master's degrees in Informatics Engineering from Universidad Técnica Federico Santa María (UTFSM) in Chile.

My research interest is focused on Human Centered Data Science and Data Visualization. I'm interested in building tools to help researchers with qualitative and automated analysis of large datasets, supporting discussions to improve inter-rater agreement by creating powerful data visualizations, inter-rater reliability metrics, and algorithms to uncover important information in the process.

• Generalized Cohen’s Kappa: a new Metric for Inter-rater Reliability Assessment for Collaborative Coding
  Advisor: Cecilia Aragon; January 2020 - Ongoing
  
  Collaborative coding of large datasets of short texts, such as tweets or comments, has been a valuable tool for qualitative research for years. When it comes to assessing the inter-rater reliability of multiple coders, existing metrics have not evolved to fit current data and present a variety of restrictions. We propose a generalized Cohen’s Kappa based on Monte Carlo simulations that can be applied to large datasets with multiple coders and non mutually exclusive categories. Preliminar results show that this new metric can replace the widely used Cohen’s kappa metric, obtaining similar results in small and restricted settings and allowing flexibility in larger settings with complex data.
  
  
• Text-prizm 2.0 - Advisor: Cecilia Aragon; June 2018 - Ongoing
  
  Qualitative coding is a labor-intensive process of manually reading and interpreting large amounts of data, many researchers use tools like Microsoft Excel or Google Spreadsheets for this task as most of the qualitative coding tools are not adapted to large online communication datasets, these tools can be effective but lack many useful features to make coding easier and more effective. Text-prizm is a web application for collaborative coding of large volumes of short text messages, its interface is built using a human-centered approach and aims to facilitate the qualitative coding and analysis of large online communication datasets.
  
  
• Cultural Differences in Data Privacy Perspectives on Social Media - Advisors: Cecilia Aragon, Claudia López; Sept 2018 - Dec 2019
  
  The Cambridge Analytica scandal has triggered a discussion about data privacy in social media. Motivated by this context, we aim to answer this research question: Does the public online debate reveal different perspectives on data privacy across countries/cultures? A large-scale Twitter dataset around this issue with both English and Spanish tweets has been collected and we aim to analyze the data through both qualitative coding and automated analysis.
  
  

• Coding: C/C++, Python, Ruby, Javascript, PHP, R
• Web Development: Ruby On Rails, JS/Jquery, HTML/CSS
• Data Visualization: Tableau, D3, Plotly, Altair, Matplotlib
• Data Manipulation: Tableau Prep, Pandas, NumPy, NLTK
• Back-end: GNU/Linux, Nginx, MariaDB, PostgreSQL, Git, Scripting/Bash
• Design: LaTeX, Adobe Photoshop, Gimp

  Download my Résumé