ELDORA Machine Learning Quality Control Dataset

The repository contains data from the airborne radar quality control process for the ELDORA airborne radar discussed in the manuscript "Airborne Radar Quality Control with Machine Learning" by Alexander J. DesRosiers and Michael M. Bell. The main data files are the '.h5' files which should be stored in a directory titled 'h5_files' locally to try and run the code. The 'make_rf_model.py' script uses the h5 data to train and evaluate a random forest model for radar quality control. The 'utils.py' script contains helper scripts needed to mine data, train, and evaluate models with this methodology which Bruno Melli helped originally develop. The 'rf_ophelia_raw_cfrad_clean.py' script can be used to clean a sample cfradial file included which should be stored in a 'sample_cfrad' directory locally. The model itself is stored in a pickle file titled 'rf_Eldo_model.pkl'.

本仓库包含Alexander J. DesRosiers与Michael M. Bell所著手稿《Airborne Radar Quality Control with Machine Learning》中讨论的ELDORA机载雷达（ELDORA airborne radar）的机载雷达质量控制流程相关数据。核心数据文件为.h5格式文件，若需运行配套代码，需将其本地存储于名为`h5_files`的目录中。`make_rf_model.py`脚本会利用上述.h5格式数据，训练并评估用于雷达质量控制的随机森林模型（Random Forest Model）。`utils.py`脚本包含基于该方法开展数据挖掘、模型训练与评估所需的辅助脚本，该方法最初由Bruno Melli参与开发。`rf_ophelia_raw_cfrad_clean.py`脚本可用于清理附带的CFRadial格式样本文件（CFRadial file），该样本文件需本地存储于`sample_cfrad`目录中。训练完成的模型存储于名为`rf_Eldo_model.pkl`的pickle格式文件（Pickle File）中。