DoViz: Intervention-centric interactive visualization for causal inference

Causal inference is the process of understanding and quantifying cause-and-effect relationships from observed data. The process of causal inference often requires analysts to use visualizations for evaluating accuracy. However, existing visualization tools often lack the ability to communicate the effect of performing interventions on the data and comparatively visualizing their results. In this thesis, we address this gap using DoViz, an intervention-centric interactive visualization prototype. DoViz enables users to interact with independent variables, simulate explicit interventions, condition on confounders, and visualize the resultant causal effects. The user interface comprises an Intervention Panel for performing intuitive manipulations and an Inference Panel for visualizing intervention outcomes. A key feature of DoViz is its ability to perform comparative causal scenario analysis. Users can juxtapose multiple scenarios side-by-side, helping them draw more informed conclusions. The design process was driven by an iterative needfinding study, distilling a user-centric workflow around intervention, analysis, and comparison.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2024

Author dc:creator

• Chandratre, Atharv Shripad

Contributors dc:contributor

• Sundaram, Hari

• Causal Inference
• Interactive Visualization
• Data Visualization Tools
• Intervention Analysis
• Comparative Scenario Analysis
• User Interface Design
• Human-computer Interaction
• Causal Reasoning
• Visualization Software

Statement dc:rights

• Copyright 2024 Atharv Chandratre

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/124696

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F124696