How hanlife02/social-science-quantitative-methods works
Project Type and Purpose
• This is a Python-based data-analysis pipeline for a social-science research project. It ingests raw CSV data, processes it, computes descriptive statistics, fits statistical models, interprets results, and produces visualizations and markdown reports. There is no frontend or web layer.
File-Structure Analysis
• Top level:
• main.py – orchestration entrypoint
• config.py – global parameters (paths, model settings)
• requirements.txt – external Python libraries (pandas, numpy, statsmodels, matplotlib, seaborn, etc.)
• data/ethnic_conflict_data.csv – raw data
• output/ – directory to hold figures and results in subfolders
• src/ – Python modules by functionality:
• data_processing.py
• descriptive_stats.py
• modeling.py
• interpretation.py
• visualization.py
Key Components
a. Data Input Layer
• Raw CSV (data/ethnic_conflict_data.csv)
• Configuration (config.py)
b. Processing & Analysis Modules
• Data Processing Module (src/data_processing.py)
• Descriptive Statistics Module (src/descriptive_stats.py)
• Modeling Module (src/modeling.py)
• Interpretation Module (src/interpretation.py)
• Visualization Module (src/visualization.py)
c. Orchestration
• main.py ties together modules in sequence, reading config values
d. Output Layer
• Figures (output/figures/*)
• Text Results and Markdown Reports (output/results/*.md)
e. External Dependencies
• Python runtime
• Libraries (pandas, numpy, statsmodels, matplotlib, seaborn, etc.)
Relationships & Data Flow
• main.py → reads config.py → loads raw CSV
• raw data → data_processing → cleaned/augmented DataFrame
• processed data → descriptive_stats → summary tables
• processed data → modeling → fitted models & coefficients
• models & stats → interpretation → narrative insights (markdown)
• processed data + model outputs → visualization → plots saved under output/figures
• All outputs collected under output/ directory
• Dependencies: each module imports pandas/numpy/statsmodels, config.py
Architectural Patterns & Principles
• Pipeline pattern: sequential stages, single responsibility per module
• Layered grouping: Input → Processing → Analysis → Presentation → Storage
• Modular design: each src/*.py encapsulates one stage, enabling testing and reuse
• Config-driven: behavior controlled by a single config.py
Technologies & Frameworks
• Python 3.x
• pandas for data ingestion & manipulation
• numpy for numeric operations
• statsmodels (or scikit-learn) for regression modeling
• matplotlib / seaborn for plotting
• markdown for narrative output
Essential Elements
• Clear textual labels for each node (module names, file paths)
• Data stores
• Code modules
• Outputs
• External libraries
• Optional: annotate module inputs/outputs (e.g., “DataFrame with X variables”)