通信情報集約

gistfile1.txt

#ファイルを分割せずに、集約して出力する
#PC間の通信回数を抽出
device_weight = []
from_device = []
to_device = []
from_packets = []
to_packets = []
from_bytes = []
to_bytes = []
duration = []

start = time.time()
index = 1
for src, sub_df in df.groupby('SrcDevice'):
    for dst, subsub_df in sub_df.groupby('DstDevice'):
        from_device.append(src)
        to_device.append(dst)
        cnt = len(subsub_df)
        device_weight.append(cnt)
        from_packets.append(subsub_df['SrcPackets'].sum())
        to_packets.append(subsub_df['DstPackets'].sum())
        from_bytes.append(subsub_df['SrcBytes'].sum())
        to_bytes.append(subsub_df['DstBytes'].sum())
        duration.append(subsub_df['Duration'].sum())
        
    print('{0} %完了'.format(index / len(SrcDevice_list) * 100))
    index += 1
#    if index > 30:
#        break
    
df_sum = pd.DataFrame({'from': from_device, 'to': to_device, 'count' : device_weight, 'duration': duration, 'from_packets': from_packets, 'to_packets': to_packets, 'from_bytes': from_bytes, 'to_bytes': to_bytes})
df_sum.to_csv('netflow_day-02_sum.csv')
elapsed_time = time.time() - start
print ("calclation_time:{0}".format(elapsed_time) + "[sec]")
df_sum