Getting packets into a dataset is only half the job. You also need useful features, consistent labels, and a way to tell whether a model learned anything.
Netcap v0.6 puts more of that work into the research workflow. It is the release used for the project's master's thesis experiments.
More useful connection records
Connection records gain TCP flag counts and mean window size. Alongside duration and bytes sent in each direction, these give an experiment more to work with than addresses and ports.
The analysis scripts also improve dataset preparation and evaluation: min–max normalization, configurable class weighting, confusion matrices, and ROC plots. These help you inspect how a model behaves when normal and attack traffic are unevenly represented.
If you're building a network-security dataset, v0.6 is a useful point in Netcap's history: capture, label, prepare, and evaluate the data in one research workflow.