Empirical Research Template

易翊翼
20241011

___________________________________________________________________________________________________________________________

Project Description

This is an empirical research project template designed to provide a standardized project structure for:

1. Version control
2. Code synchronization between local machine and HPC
3. Decoupling of code and data
4. Separation of build and analysis phases

The template structure follows the Guide by Matthew Gentzkow and Jesse M. Shapiro and includes useful config.do and config.py files for setting up paths and packages in Stata and Python.

Directory Structure

├── analysis            # Analysis phase
│   ├── code            # Analysis code
│   └── data
│       ├── input←←←←←| # Panel data for descriptive stats and regressions(4) 
|       |    ↓        |
│       └── output    | # Generated tables and figures (5)
├── build             | # Data construction phase
│   ├── code          | # Data processing code
│   └── data          |
│       ├── raw       | # Raw data (1) 
|       |    ↓        | 
│       ├── temp      | # Temporary files, merge keys, etc. (2) 
|       |    ↓        |
│       └── processed→| # Processed databases (3) 
├── README.md           # Project documentation
├── README.py           # Directory tree generator
└── resource            # Related papers and materials

Usage Guide

1. Data Management

• build/data/raw: Store raw data (read-only)
• build/data/processed: Store processed data
• build/data/temp: Store intermediate files
• analysis/data/input: Store analysis-ready data
• analysis/data/output: Store analysis results

2. Code Organization

• build/code: Data cleaning and construction code
• analysis/code: Analysis code
• Each code file should have clear documentation

3. Version Control

• Use .gitignore for large data files
• Use .gitkeep to maintain empty directories
• Regular code commits

4. Configuration

• Use config.py for Python paths and packages
• Use config.do for Stata paths and packages

Best Practices

1. Data Security

   • Don't commit sensitive data
   • Use .gitignore for large files

2. Code Standards

   • Keep code clean and documented
   • Use meaningful names
   • Add appropriate comments

3. Performance

   • Use chunking for large datasets
   • Choose appropriate data structures

4. Collaboration

   • Regular code sync
   • Keep documentation updated
   • Follow project standards

Maintenance

• Regular dependency updates
• Documentation maintenance
• Temporary file cleanup
• Data backup