ValueError: Expect x to be a non-empty array or dataset

[vinhqdang]

Hi there

First of all, many thanks for publishing the code.

I am trying to re-run your code (TensorFlow version).

What I did:

• install latest tensorflow (CPU version) 2.5
• numpy 1.19 (otherwise there will be an error of feeding tensorflow to numpy).
• networkx 2.6.2
• matplotlib

I downloaded the CICIDS 2017, unzip the file GeneratedLabelledFlows.zip and get the list of 8 CSV files.

For each CSV file, I sort them by Timestamp column, then store first 70% of lines to train.csv and last 30% to val.csv

The code:

for i,f in tqdm (enumerate (csv_files)):
    fname = join ('TrafficLabelling', f)
    df = pd.read_csv (fname, low_memory=False)
    df = df.sort_values (by = ' Timestamp')
    N = df.shape [0]
    train_filename = 'train/' + f
    val_filename = 'val/' + f
    df_train = df[0:int(N*.7)]
    df_val = df[int(N*.7):]
    df_train.to_csv (train_filename, index = False)
    df_val.to_csv (val_filename, index = False)

Then I modify the config.ini and run python main.py (Python 3.8.11)

I got the following error:

raise ValueError('Expect x to be a non-empty array or dataset.')
ValueError: Expect x to be a non-empty array or dataset.

Do you have any idea what did I do wrong?

Many thanks

[justussoh]

@vinhqdang Did you manage to solve the issue?

[vinhqdang]

hi @justussoh , no, I did not

[otmanjai]

@vinhqdang & @justussoh Did you manage to solve the issue?