I have to predict number of packets in my network traffic for each IP address independently. I have 8 IP adressess in my dataset, timewindow, isSending flag, weekday (X) and number of packets (Y). I prepared one-hot encoding for every columns. My aim is to make traffic prediction for given vector: IP, day, timewindow, isSending.
I tried MLPRegressor treating whole dataset as one. It's giving a false output now because it treats whole data as one 'class'. I would like to prepare line regression for each ip address separately, but i have such a big dataset that it is impossible to fit model separately on my own.
Example of row from my dataset : 0;0;0;0;0;0;0;0;1;1;0;0;0;0;0;0;0;1;9823 Contains 1 on possitions corresponding source IP, timewindow id, day. It's a dataset just for test with only several timewindows and 1 day. At the end is number of packets that flowed from IP in certain timewindow and day.
My outputs are tottaly unacceptable, for example i expect 3 and have -134
