Training accuracy is 0.0%

[C1PO]

I've already read the old issue but I think my pcap are larger than them.
How can I solve it?

OUT: 71,IN: 75,TOTAL: 146,SIZE: 36082,RATIO: 1.056338028169014
    1 pcap files
 - google.com
OUT: 200,IN: 207,TOTAL: 407,SIZE: 110322,RATIO: 1.035
    1 pcap files
 - android.com
OUT: 106,IN: 164,TOTAL: 270,SIZE: 195666,RATIO: 1.5471698113207548
OUT: 1207,IN: 1903,TOTAL: 3110,SIZE: 2497937,RATIO: 1.5766362883181442
OUT: 119,IN: 117,TOTAL: 236,SIZE: 65432,RATIO: 0.9831932773109243
OUT: 60,IN: 69,TOTAL: 129,SIZE: 73283,RATIO: 1.15
    4 pcap files
 - facebook.com
OUT: 836,IN: 1290,TOTAL: 2126,SIZE: 1576638,RATIO: 1.5430622009569377
OUT: 50,IN: 103,TOTAL: 153,SIZE: 20293,RATIO: 2.06
OUT: 432,IN: 473,TOTAL: 905,SIZE: 398269,RATIO: 1.0949074074074074
OUT: 87,IN: 92,TOTAL: 179,SIZE: 61704,RATIO: 1.0574712643678161
    4 pcap files
 - volkswagen.it
OUT: 128,IN: 163,TOTAL: 291,SIZE: 152680,RATIO: 1.2734375
OUT: 941,IN: 1428,TOTAL: 2369,SIZE: 1789311,RATIO: 1.5175345377258236
OUT: 107,IN: 152,TOTAL: 259,SIZE: 158090,RATIO: 1.4205607476635513
OUT: 116,IN: 145,TOTAL: 261,SIZE: 152095,RATIO: 1.25
    4 pcap files
 - wikipedia.org
OUT: 65,IN: 62,TOTAL: 127,SIZE: 47268,RATIO: 0.9538461538461539
OUT: 109,IN: 120,TOTAL: 229,SIZE: 103199,RATIO: 1.1009174311926606
    2 pcap files
 - duckduckgo.com
OUT: 78,IN: 79,TOTAL: 157,SIZE: 47383,RATIO: 1.0128205128205128
    1 pcap files
Training size: 15
Testing size:  2
Accuracy: 0.0%

[wisepythagoras]

Hi, sorry @C1PO . I just saw this. I don't get any notifications for some reason. Even though the pcaps are larger, the model is only trained on a certain amount of packets.

This might happen because of two reasons:

Either you don't have a large enough dataset, and/or there is too much noise in the pcap files. Try to only scrape packets to and from the website you are trying to fingerprint. For example, only open one tab to that website. After the first load stop the recording and repeat so you can collect more data.

If I get a chance, I can try to do this as well tomorrow and share the model with you