flow_models.elephants package

flow_models.elephants.plot_entropy module

Generates plot of features entropy and (optionally) importances.

calculate_entropy(directory)

Calculate entropy of 5-tuple features for given flow records directory.

Parameters:: directory (os.PathLike) – binary flow records directory
Returns:: entropy for subsequent bytes and bits of (sa, da, sp, dp, prot) fields
Return type:: dict

Simulates flow table for elephant flow classification using flow records data directory.

simulate_data(directory, index=Ellipsis, mask=None, pps=None, fps=None, timeout=15, max_seconds=3600)

Simulate flow table occupancy reduction curve for given flow records.

Parameters:

directory (os.PathLike) – binary flow records directory
index (np.array, default Ellipsis) – index array of flows to use for simulating data
mask (np.array[bool], optional) – flows to add to flow table (elephants)
pps (float, optional) – packets per second, when None flow records times are used for calculation
fps (float, optional) – flows per second, when None flow records times are used for calculation
timeout (float, default 15.0) – inactive flow table timeout in seconds
max_seconds (int, default 3600) – total seconds number of simulation

Returns:

flows_sum (int) – sum of flows added to flow table
octets_sum (int) – sum of octets transmitted by flows while being in flow table
flows_slots (np.array) – number of flows present in flow table in each second
octets_slots (np.array) – amount of octets trasmitted by flows in flow table in each second