proj_pcap

Idea: classify traffic on anomaly/normal. If anomlay -> classify further approach: generalization by machine learning

MainFlow:

Dataset creation: labeled.pcap -> labeled.pcap_splitted by 1 second 
			       -> creating csv file with network packet description on each of splitted
			       -> feature extraction (for example - simple statistic calculation)
			       -> adding each record to TRAIN dataset
			       --> here should be evaluation (not performed)

Model creation: using created labeled dataset:
			       -->choose model (not performed)
			       --> feature evaluation (basic evaluation perfomred) //conclusion not enough data -> delete stupid correlations
			       -> train logisticRegression anyway, perform crossvalidation.
			       -> save model as binary object using pickle or save weights of model if can be done (ex. lightgbm)
    Assemble Program (PROJ):
                                   -> copy from to
			       -> copy Model_creation/binary PROJ_example/funs/model_data
			       -> copy Model_creation/multi PROJ_example/funs/model_data
			       -> copy Model_creation/available_cat.txt PROJ_example/funs
			       -> run program on machine with tcpdump

Folders (all code, with comments):

Dataset_creation # modify wireshark/editcap path in Dataset_creation/split.py
Model_creation  
Program

Basic requirements: for WM ubuntu/debian python3.8 for others: ptyhon3.6 or higher (should be mine works fine)

pip install scapy tqdm pandas numpy sklearn (can be deleted manualy)

Additionally: wireshark editcap util