-1
$\begingroup$

i am working on my final year project that is a social network. Based on user interest, i have to add him in groups based on area he lives in, age group, interest type, gender and some other features.i have to use machine learning to predict in which group should i place him. I am thinking of using classification but i need data set to train which i don't have at all nor there is any data set related to this problem, so if any one of you guide me in right direction how can i do this ?

$\endgroup$
  • $\begingroup$ a heuristic approach would easily classify users based on area he lives in, age group, interest type, and gender. Depending on the other features, is ML even appropriate? $\endgroup$ – Hobbes Apr 25 '17 at 19:08
  • $\begingroup$ What data do you have? $\endgroup$ – emilliman5 Apr 25 '17 at 19:58
  • $\begingroup$ actually i have no idea if ML should be used. i am explaining my problem to you, hopefully you will guide me to right direction. User registers and add his interests like hiking, swimming, football, karate, programming etc, now based on his interest, the area he lives in, age group and some other features, i have to add him in a specific group, i was doing it in traditional way but my professor said there should be intelligence in system so that system intelligently add him to a specific group. now what should i do ? $\endgroup$ – Ahmad Raza Apr 26 '17 at 5:06
1
$\begingroup$

As far as I know you shouldn't use ML at this stage. There are two problems:

  1. You can not get enough data related to your task.
  2. Even anyhow you manage to get the data it will be either to general or
    will be some other domain based so relying on the result would be difficult.

Instead you can write some rule based or weighted value solutions like depending upon the type of group what age-group should be part of the group. Regular expressions to extract some keywords from interest tags and other text features and these can be used to decide the probability of a person falling into a group.

Once if you have enough data and their labels using this method you can start trying ML algorithms and they will give the results.

$\endgroup$
  • $\begingroup$ i have a little solution that i thought of, lets assume i have data that is all correct, like for which particular person, a group should be suggested. a new user comes and we match his data with current data to suggest which group he lies in, is this kind of machine intelligence or not ? $\endgroup$ – Ahmad Raza Apr 26 '17 at 11:57
  • $\begingroup$ Machine Intelligence is something entirely different. ya it is kind of statistical model which decides according to degree of match. $\endgroup$ – Prayalankar Apr 28 '17 at 12:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.