Replication Data and Code for: Human-Network Regions as Effective Geographic Units for Disease Mitigation

This repository contains code and partial data for replicating the work performed in "Human-Network Regions as Effective Geographic Units for Disease Mitigation" as well as an online resource for interacting with regional boundaries. Please contact corresponding author Clio Andris for assistance with implementing code.<br><br>An interactive webtool accompanying the paper "Human-Network Regions as Effective Geographic Units for Disease Mitigation" allows users to create regions based on a combination of weighted inputs from different network data. Each county is then colored by region. Users can export the resultant regions as a tabular file, and export an image of the map (e.g., jpg or tif). The webtool is here: https://www.geo-social.com/consensus_regions/index.html<br><br>FILE DESCRIPTIONSCountyFIPS_Communities.csv: This dataset contains community detection results for each U.S. county in the Conterminous U.S., where the community detection methods divide counties into separate groups (i.e., clusters). There are five different input networks (Commuters, Facebook relationships, Migrants, GPS-based trips and Twitter co-mentions). The different methods were selected using the igraph package in R. A Q statistic is reported for each network/method combination.Migrants_Commuters.csv: This dataset is an undirected edgelist of counties in the U.S. weighted by number of migrants and number of commuters between the counties. Each county is given using a FIPS number.<br><br>nullmodel1000_44.csv: This dataset provides 'null' (i.e., randomized) regions at the U.S. county level. Each county is assigned to one of 44 different regions using random points and Voronoi polygons. Each region is geographically contiguous. There are 1,000 randomized permutations provided.Lattice2.csv: This dataset is a county adjacency edgelist by FIPS code.<br><br>R Code (Steps 1 - 4): This is the R code used to import, clean, and analyze the data.ConsensusRegions_Webtool.txt: Lists the website for our online webtool (https://www.geo-social.com/consensus_regions/index.html).<br>

本仓库包含用于复现《Human-Network Regions as Effective Geographic Units for Disease Mitigation》研究工作的代码与部分数据，同时提供用于交互浏览区域边界的在线资源。如需获取代码实现相关协助，请联系通讯作者Clio Andris。

本论文配套的交互式网页工具《Human-Network Regions as Effective Geographic Units for Disease Mitigation》允许用户基于多类网络数据的加权输入组合生成区域，随后将按区域为各县级行政区着色。用户可将生成的区域导出为表格文件，并导出地图图像（如JPEG（JPEG）、TIFF（TIFF）格式）。该网页工具地址为：https://www.geo-social.com/consensus_regions/index.html

## 文件说明
CountyFIPS_Communities.csv：本数据集包含美国本土连续区域（Conterminous U.S.）内各县级行政区的社区检测结果，其中社区检测方法将县级行政区划分为若干独立群组（即簇）。本次实验共采用5类不同的输入网络：通勤网络（Commuters）、Facebook（Facebook）社交关系网络、移民网络（Migrants）、基于GPS的出行网络以及Twitter联合提及网络。所有检测方法均通过R语言的igraph包（igraph）实现，且为每一类网络/方法组合报告了Q统计量。

Migrants_Commuters.csv：本数据集为美国县级行政区的无向边列表，边权重由县际移民数量与通勤数量决定，各县级行政区均以FIPS编码（FIPS）标识。

nullmodel1000_44.csv：本数据集提供美国县级行政区层面的空模型（null model）区域结果，即通过随机点与沃罗诺伊多边形（Voronoi polygons）方法，将各县级行政区分配至44个地理连续的区域中。本次共提供1000组随机置换结果。

Lattice2.csv：本数据集为基于FIPS编码的县级行政区邻接边列表。

R代码（步骤1-4）：用于导入、清洗与分析本数据集的R代码文件。

ConsensusRegions_Webtool.txt：列明本在线网页工具的网址（https://www.geo-social.com/consensus_regions/index.html）。