Pharos Data

IntroductionThis data repository includes the cleaned dataset for the Pharos application 2023-2024 data collection period (May 2023-March 2024). This dataset includes the full recurring network measurement (RNM), landmark (LM) datasets, as well as the county geographies used for the study catchment area. Also included in this dataset are the necessary software files to clean and visualize the collected data replicating the methods used in our published analysis.Figures Generated by This RepositoryRunning the Python scripts in this repository generates the following manuscript figures:<b>Figure 2</b> - Upload/download success counts and duration distributions by carrier<b>Figure 5</b> (top panel) - Count of datapoints per county (log scale)<b>Figure 5</b> (bottom panel) - Temporal trends for Kisumu county (radar plots by carrier)<b>Figure S2A</b> - Temporal trends for Siaya county<b>Figure S2B</b> - Temporal trends for Migori county<b>Figure S2C</b> - Temporal trends for Homa Bay countySetup and Execution Instructions for ReproductionPrerequisites<b>Python 3.9.16</b> (likely compatible, but untested with Python &gt;3.7)<b>pip</b> (Python package installer)Files Included<b>_main.py</b> - Main execution script<b>_clean_df.py</b> - Data cleaning module<b>_make_viz.py</b> - Visualization module<b>_clean_</b><b>lms.csv</b> - Landmark data<b>_clean</b><b>_</b><b></b><b>rnms</b><b>.csv</b> - Recurring network measurement data<b>_Counties</b><b>_WesternKenya.json</b> - Geographic boundaries for Western Kenya counties<b>requirements.txt</b> - Python package dependenciesArcGIS Figures Reproducibility Steps.pdf/.docx - Instructions for reproducing ArcGIS-generated figuresInstallation and ExecutionCreate a virtual environment (recommended, but not required)<br><code>python3 -m venv venv</code>Activate the virtual environment (recommended, but not required)<br><code>source venv/bin/activate</code>Install required packages<br><code>pip install -r requirements.txt</code>Run the analysis<br><code>python _main.py</code><br>The script will generate PNG files for <b>Figures 2</b>, <b>5</b>, <b>S2A</b>, <b>S2B</b>, and <b>S2C</b> at 300 DPI resolution, suitable for publication.Deactivate virtual environment when done (if used)<br><code>deactivate</code><b>Reproducing Other Manuscript Figures</b>Other figures in the manuscript (<b>Figure 4</b>, <b>Figure</b><b> 6</b>, <b>Figure</b><b> 7</b>, <b>Figure S</b><b>1</b>, <b>Figure</b><b> S3</b>, and <b>Figure</b><b> S4</b>) were created using ArcGIS Pro with the same RNM and LM datasets included in this repository. Step-by-step instructions for reproducing these ArcGIS-generated figures are provided in the included files: <b>ArcGIS Figures Reproducibility Steps.pdf</b> and <b>ArcGIS Figures Reproducibility Steps.docx</b>.

一、引言
本数据集仓库涵盖了Pharos应用程序2023-2024年度（2023年5月至2024年3月）数据采集阶段的清洗后数据集。本数据集包含完整的周期性网络测量（Recurring Network Measurement, RNM）、地标（Landmark, LM）数据集，以及本研究覆盖区域所用的县域地理边界数据。本数据集同时附带了用于清洗、可视化采集数据的必要软件文件，可复现我们已发表分析中所采用的研究方法。

二、本仓库可生成的图表
运行本仓库内的Python脚本可生成以下论文配图：<b>图2</b>——按运营商划分的上传/下载成功次数与时长分布；<b>图5</b>（上半部分）——各县域数据点数量（对数刻度）；<b>图5</b>（下半部分）——基苏木县的时间趋势（按运营商绘制的雷达图）；<b>补充图S2A</b>——西亚亚县的时间趋势；<b>补充图S2B</b>——米格奥里县的时间趋势；<b>补充图S2C</b>——霍马贝县的时间趋势。

三、复现实验的设置与执行说明
1. 前置依赖
<b>Python 3.9.16</b>（大概率兼容Python 3.7及以上版本，但未针对Python >3.7进行测试）；<b>pip</b>（Python包安装器）。

2. 包含文件
<b>_main.py</b>——主执行脚本；<b>_clean_df.py</b>——数据清洗模块；<b>_make_viz.py</b>——可视化模块；<b>_clean_lms.csv</b>——地标数据集；<b>_clean_rnms.csv</b>——周期性网络测量数据集；<b>_Counties_WesternKenya.json</b>——肯尼亚西部各县的地理边界数据；<b>requirements.txt</b>——Python包依赖列表；<b>ArcGIS Figures Reproducibility Steps.pdf</b> / <b>ArcGIS Figures Reproducibility Steps.docx</b>——复现ArcGIS生成图表的操作指南。

3. 安装与执行步骤
创建虚拟环境（推荐，非强制）：<br><code>python3 -m venv venv</code>
激活虚拟环境（推荐，非强制）：<br><code>source venv/bin/activate</code>
安装所需依赖包：<br><code>pip install -r requirements.txt</code>
运行分析脚本：<br><code>python _main.py</code>
该脚本将以300 DPI分辨率生成<b>图2</b>、<b>图5</b>、<b>补充图S2A</b>、<b>补充图S2B</b>与<b>补充图S2C</b>的PNG格式文件，符合出版级要求。
运行结束后退出虚拟环境（若已创建）：<br><code>deactivate</code>

四、复现其余论文配图
本论文其余配图（<b>图4</b>、<b>图6</b>、<b>图7</b>、<b>补充图S1</b>、<b>补充图S3</b>与<b>补充图S4</b>）均使用ArcGIS Pro制作，所用RNM与LM数据集与本仓库内含数据集一致。复现这些ArcGIS生成图表的分步操作指南已随附于以下文件中：<b>ArcGIS Figures Reproducibility Steps.pdf</b>与<b>ArcGIS Figures Reproducibility Steps.docx</b>。