RDS Tools Documentation

RDS Tools is a Python package for Respondent-Driven Sampling (RDS) analysis and bootstrap resampling with parallel processing capabilities.

Contents

User Guide

Installation

cd RDSTools
pip install .

For development:

pip install -e .

Quick Start

from RDSTools import load_toy_data
from RDSTools import RDSdata, RDSmean, RDStable, RDSlm

# Load the included example dataset
toy_data = load_toy_data()

# Process RDS data
# Or load your own data:
# import pandas as pd
# data = pd.read_csv("survey_data.csv")
rds_data = RDSdata(
    data=toy_data,
    unique_id="ID",
    redeemed_coupon="CouponR",
    issued_coupons=["Coupon1", "Coupon2", "Coupon3"],
    degree="Degree"
)

# Calculate means with bootstrap variance
result = RDSmean(
    x='Age',
    data=rds_data,
    weight='WEIGHT',
    var_est='tree_uni1',
    resample_n=1000,
    n_cores=4  # parallel processing
)
print(result)

# Create frequency tables
table = RDStable(
    x='Sex',
    data=rds_data,
    weight='WEIGHT',
    var_est='tree_uni1',
    resample_n=1000
)
print(table)

# Fit regression models
model = RDSlm(
    data=rds_data,
    formula='Income ~ Age + C(Sex) + C(Race)',
    weight='WEIGHT',
    var_est='tree_uni1',
    resample_n=1000,
    n_cores=4
)
print(model)

Key Features

Data Processing

  • RDSdata() - Process RDS survey data and create network structure

  • Automatic wave detection and seed identification

  • Flexible degree imputation methods (mean, median, hotdeck, drop)

Estimation

  • RDSmean() - Calculate means with RDS-adjusted standard errors

  • RDStable() - Generate one-way and two-way frequency tables

  • RDSlm() - Linear and logistic regression models

  • Weighted and unweighted analyses

  • Naive and bootstrap variance estimation

Bootstrap Variance

  • RDSboot() - Six bootstrap resampling methods

  • RDSBootOptimizedParallel() - Parallel bootstrap for performance

  • Chain, tree unidirectional, and tree bidirectional methods

Visualization

  • RDSnetgraph() - Network visualizations with multiple layouts

  • RDSmap() - Interactive geographic maps

  • Helper functions: get_available_seeds(), get_available_waves(), print_map_info()

Performance

  • Parallel processing support with n_cores parameter

  • Up to 10x speedup with 8 cores

  • Optimized algorithms for large datasets

Indices and Tables