Need to convert main.cpp to Python. The main function reads the sample dataset and then runs the anomaly detection algorithm on it, reporting the accuracy.

To improve the performance of data loading, we should try to port the read_data function to Rust, which will run a lot faster (since it doesn't have to allocate so many small objects). This issue outlines how we could do it.

From a code perspective:

• Open and read the file indicated by the given path (up to the limit of rows we're told to read).
• Create a new Rust structure for storing each TimeEdge.
• Store the edges in a vector.
• Perform the operations the existing Python code does (find min/max of node indices, subtract them, etc.)
• Convert the result to numpy format using rust-numpy and then return it back to Python.

Besides the code, we should also:

• convert the Python docstring to Rust doc comments, which will be automatically converted to docstrings by pyo3
• port the unit tests to Rust

compute_anomaly_score calculates how suspect is the given network traffic.

Need to convert pagerank.cpp in Python.

This is the Wikipedia article for PageRank.

To ensure the code works, we should create a unit and integration test suite, automated with pytest.

Initially, we could keep tests alongside the library code.

The layout would be:

anomrank/
 - __init__.py
 - anomaly_detect.py
 - read_data.py
 - test/
    - __init__.py
    - test_anomaly_detect.py
    - test_read_data.py
    - data.txt

• test_anomaly_detect.py: runs the functions in anomaly_detect.py
• test_read_data.py: runs the functions in read_data.py.
• data.txt: small file in which we place some values for the read_data function

accuracy.cpp tells how good it is at detecting anomalies over a dataset.

We can use this function. Maybe it helps.

The read_data function reads into memory the list of edges, with their corresponding time stamps.

I've forked the original code and updated it to use modern C++ features in this repo.

Details

Need to convert the C++ files from here to Python code.

Components

• main

• read_data

• PageRank

• accuracy

• anomaly_detect

• anomaly_inject

The function inject adds a new edge on the graph between two nodes. inject_anomaly adds an anomaly to the graph and inject_snapshot adds the current state of the graph.