How oscarski/research-workshop-cognitive-science-and-ai works
Project Type and Purpose
• This is a data‐analysis pipeline focused on survey data for “Virtual Agent Perception Analysis.”
• It ingests raw CSVs, cleans and prepares the data, runs statistical models, and produces visualizations and a cleaned dataset.
File Structure Analysis
• Top level:
• data_cleaning.py – ETL/cleaning routines
• data_analysis.py – statistical modeling and orchestration
• cleaned_data.csv – processed output
• Version 1–4 .csv files – raw survey inputs
• Subdirectory (per README):
• Visualizations/ – charts and figures
• No frontend/backend split – it’s a single‐machine pipeline in Python.
Key Components
a. Data Sources
• Version1.csv, Version2.csv, Version3.csv, Version4.csv
b. Data Cleaning Module
• data_cleaning.py
• tasks: load raw, handle missing, cast types, create participant IDs, filter
c. Cleaned Data Store
• cleaned_data.csv
d. Analysis Module
• data_analysis.py
• tasks: compute composite scores, run logistic & ordinal regressions, descriptive stats
e. Visualization Module
• uses matplotlib & seaborn to produce .png files in Visualizations/
f. Outputs
• cleaned_data.csv
• composite_scores.png, demographics.png, rq1_consistency_persuasion.png
Relationships & Data Flow
• Cleaning Module → cleaned_data.csv
• cleaned_data.csv → Analysis Module
• Analysis Module → Visualization Module
• Visualization Module → output .png files
Architectural Patterns & Principles
• Single responsibility per script
• File‐based communication (CSV + PNG artifacts)
• Modular design: cleaning vs. analysis
Technologies & Frameworks
• Python 3.x
• pandas, numpy for data handling
• matplotlib, seaborn for viz
• statsmodels & scipy for statistical models