ML Preprocessing Scripts for Numerical Array Conversion

This repository by sheatsley provides Python 3 scripts for downloading and converting 10+ machine learning datasets into NumPy arrays, with the most recent update in January 2025. It focuses on automating the preprocessing pipeline for specific benchmarks in cybersecurity and computer vision, such as CIC-MalMem-2022 and CIFAR-10.

Use Cases

• Training network intrusion detection systems using the NSL-KDD or UNSW-NB15 preprocessing scripts
• Developing malware detection models based on the CIC-MalMem-2022 memory analysis features
• Benchmarking image classification algorithms using the CIFAR-10 and Fashion-MNIST numpy-ifiers

Strengths

• Supports 10+ distinct datasets across cybersecurity and vision domains
• Standardizes output to NumPy format for direct integration with Python ML libraries
• Released under the BSD-3-Clause license allowing for flexible redistribution

Limitations

• Requires local execution of scripts to generate data files
• Dependent on the availability of external hosting for the original source datasets
• Lacks direct documentation of feature columns within the repository metadata

Provenance

Source

sheatsley (GitHub)

Collection Method

Scripted preprocessing of publicly available benchmark datasets

Freshness

Last updated January 2, 2025.