Installation

pip install -r requirements.txt

Instructions

• all columns have been used to train the model. Test data f1-score is 99.9%

• Inside the API folder run (on CMD):

python manage.py runserver

this will start server at : http://127.0.0.1:8000

• send a get request to http://127.0.0.1:8000/predict/ (you can use postman for testing)

sample JSON request:

{ "service":"-", "state":"FIN", "proto":"tcp", "attack_cat":"Normal", "ID_NO":12, "dur":100, "spkts":10, "dpkts":10, "sbytes":100, "dbytes":100, "rate":100, "sttl":100, "dttl":100, "sload":12, "dload":12, "sloss":12, "dloss":12, "sinpkt":15, "dinpkt":15, "sjit":15, "djit":15, "swin":15, "stcpb":48, "dtcpb":48, "dwin":48, "tcprtt":98, "synack":98, "ackdat":98, "smean":98, "dmean":98, "trans_depth":8, "response_body_len":54, "ct_srv_src":56, "ct_state_ttl":55, "ct_dst_ltm":55, "ct_src_dport_ltm":87.215, "ct_dst_sport_ltm":87.215, "ct_dst_src_ltm":87.215, "is_ftp_login":87.215, "ct_ftp_cmd":87.215, "ct_flw_http_mthd":0.225, "ct_src_ltm":0.225, "ct_srv_dst":0.225, "is_sm_ips_ports":1.25 }

• API sends a response :

{ "pred_probability": 0.0, "input_request": { "service": "-", "state": "FIN", "proto": "tcp", "attack_cat": "Normal", "ID_NO": 12, "dur": 100, "spkts": 10, "dpkts": 10, "sbytes": 100, "dbytes": 100, "rate": 100, "sttl": 100, "dttl": 100, "sload": 12, "dload": 12, "sloss": 12, "dloss": 12, "sinpkt": 15, "dinpkt": 15, "sjit": 15, "djit": 15, "swin": 15, "stcpb": 48, "dtcpb": 48, "dwin": 48, "tcprtt": 98, "synack": 98, "ackdat": 98, "smean": 98, "dmean": 98, "trans_depth": 8, "response_body_len": 54, "ct_srv_src": 56, "ct_state_ttl": 55, "ct_dst_ltm": 55, "ct_src_dport_ltm": 87.215, "ct_dst_sport_ltm": 87.215, "ct_dst_src_ltm": 87.215, "is_ftp_login": 87.215, "ct_ftp_cmd": 87.215, "ct_flw_http_mthd": 0.225, "ct_src_ltm": 0.225, "ct_srv_dst": 0.225, "is_sm_ips_ports": 1.25 } }

pred_probability gives the predicted probability in range 0.0 --> 1.0.