africa-stp-views-conflict-forecasts
收藏资源简介:
该数据集名为“圣多美和普林西比 - VIEWS冲突预测”,由Violence & Impacts Early-Warning System(VIEWS)发布,旨在提供圣多美和普林西比(STP)未来三年的月度冲突预测。数据集包含36行数据,每行代表国家层面的聚合数据,共有12个列(8个数值型,4个分类型)。数据集分为训练集(28行)和测试集(7行)。主要变量包括地理标识(如country_id、isoab)、时间标识(如year、month_id)、元数据(如name、gwcode)以及冲突预测相关指标(如main_mean_ln、main_mean)。数据集适用于表格分类和回归任务,特别适用于冲突预测和人道主义研究领域。数据来源于HDX平台,经过Electric Sheep Africa整理为Parquet格式,并进行了标准化处理。
This dataset is named "São Tomé and Príncipe - VIEWS Conflict Prediction", released by the Violence & Impacts Early-Warning System (VIEWS). It aims to provide monthly conflict forecasts for São Tomé and Príncipe (STP) over the upcoming three years. The dataset contains 36 rows of national-level aggregated data, with 12 columns in total (8 numerical and 4 categorical). It is split into a training set (28 rows) and a test set (7 rows). Key variables include geographic identifiers (e.g., country_id, isoab), temporal identifiers (e.g., year, month_id), metadata (e.g., name, gwcode), and conflict prediction-related metrics (e.g., main_mean_ln, main_mean). This dataset is suitable for tabular classification and regression tasks, and is particularly applicable to the fields of conflict prediction and humanitarian research. The data is sourced from the HDX platform, formatted into Parquet format by Electric Sheep Africa, and has been standardized.
数据集概述
基本信息
- 数据集名称:Sao Tome and Principe - VIEWS conflict forecasts
- 发布者:Violence & Impacts Early-Warning System
- 数据来源:HDX (https://data.humdata.org/dataset/stp-views-conflict-forecasts)
- 许可证:cc-by-sa-4.0
- 语言:英语
- 多语言性:单语
- 数据规模:n<1K
- 任务类别:表格分类、表格回归
- 标签:africa, humanitarian, hdx, electric-sheep-africa, conflict-violence, fatalities, forecasting, hxl, stp
- 最后更新日期(HDX):2026-04-01
- ML格式处理者:Electric Sheep Africa (https://huggingface.co/electricsheepafrica)
数据集内容
- 领域:冲突与安全
- 观察单位:国家级汇总数据
- 地理范围:STP(圣多美和普林西比)
- 总行数:36
- 总列数:12(8个数值型,4个类别型,0个日期时间型)
- 数据分割:
- 训练集:28行
- 测试集:7行
变量说明
- 地理变量:
country_id(范围80.0–80.0),isoab(STP),year(范围2026.0–2029.0) - 时间变量:
month_id(范围555.0–590.0),month(范围1.0–12.0) - 标识符/元数据:
name(Sao Tome and Principe),gwcode(范围403.0–403.0),esa_source(HDX),esa_processed(2026-04-08) - 其他变量:
main_mean_ln(范围0.0045–0.1585),main_mean(范围0.0046–0.1717),main_dich(范围0.0–0.0)
数据模式
| 列名 | 类型 | 空值百分比 | 范围/示例值 |
|---|---|---|---|
country_id |
int64 | 0.0% | 80.0 – 80.0 (均值 80.0) |
month_id |
int64 | 0.0% | 555.0 – 590.0 (均值 572.5) |
name |
object | 0.0% | Sao Tome and Principe |
gwcode |
int64 | 0.0% | 403.0 – 403.0 (均值 403.0) |
isoab |
object | 0.0% | STP |
year |
int64 | 0.0% | 2026.0 – 2029.0 (均值 2027.1667) |
month |
int64 | 0.0% | 1.0 – 12.0 (均值 6.5) |
main_mean_ln |
float64 | 0.0% | 0.0045 – 0.1585 (均值 0.1075) |
main_mean |
float64 | 0.0% | 0.0046 – 0.1717 (均值 0.1149) |
main_dich |
float64 | 0.0% | 0.0 – 0.0 (均值 0.0) |
esa_source |
object | 0.0% | HDX |
esa_processed |
object | 0.0% | 2026-04-08 |
数值摘要
| 列名 | 最小值 | 最大值 | 均值 | 中位数 |
|---|---|---|---|---|
country_id |
80.0 | 80.0 | 80.0 | 80.0 |
month_id |
555.0 | 590.0 | 572.5 | 572.5 |
gwcode |
403.0 | 403.0 | 403.0 | 403.0 |
year |
2026.0 | 2029.0 | 2027.1667 | 2027.0 |
month |
1.0 | 12.0 | 6.5 | 6.5 |
main_mean_ln |
0.0045 | 0.1585 | 0.1075 | 0.1319 |
main_mean |
0.0046 | 0.1717 | 0.1149 | 0.141 |
main_dich |
0.0 | 0.0 | 0.0 | 0.0 |
数据处理
- 原始数据通过CKAN API从HDX下载并转换为Parquet格式。
- 列名被转换为小写和蛇形命名法。
- 常见的缺失值标记被统一为
NaN。 - 数据集使用固定的随机种子(42)按80/20的比例分割为训练集和测试集,并保存为Snappy压缩的Parquet文件。
使用说明
python from datasets import load_dataset ds = load_dataset("electricsheepafrica/africa-stp-views-conflict-forecasts") train = ds["train"].to_pandas() test = ds["test"].to_pandas()
局限性
- 数据来源于Violence & Impacts Early-Warning System,未经ESA独立验证。
- 自动清洗无法纠正原始收集中误报的值、定义不一致或抽样偏差。
- 有关发布者自身的方法说明和注意事项,请参考原始HDX数据集页面(https://data.humdata.org/dataset/stp-views-conflict-forecasts)。
引用
bibtex @dataset{hdx_africa_stp_views_conflict_forecasts, title = {Sao Tome and Principe - VIEWS conflict forecasts}, author = {Violence & Impacts Early-Warning System}, year = {2026}, url = {https://data.humdata.org/dataset/stp-views-conflict-forecasts}, note = {Repackaged for machine learning by Electric Sheep Africa (https://huggingface.co/electricsheepafrica)} }




