machine learning - Neural network: how to normalize user behavior as input -


i working on neural network project reads server log files , tries categorize users candidate buyers , random surfers buckets.

the log file looks like

timestamp, user_id, event, page 1454973364, user_a, enter, http://some_url/?page=dashboard 1454973365, user_a, enter, http://some_url/?page=search&q=tablet 1454973366, user_a, enter, http://some_url/?page=faq&item=1234 1454973366, user_b, enter, http://some_url/?page=about_us 1454973366, user_a, enter, http://some_url/?page=order_placed ... 1454973368, user_a, exit, http://some_url/?page=order_placed 1454973368, user_b, exit, http://some_url/?page=about_us 

all users user_a should positively targeted candidate buyers , user_b should negatively targeted , categorized random surfers. of course, goal of project make prediction before user hits conversion page order_placed.

  1. what best way model series of events inputs?

  2. a more general question. log data can obtain info how users engage our site (drag widget, click link, scroll down/up, time spent on page, etc.). given array of actions these, general approach in machine learning model them inputs?


Comments

Popular posts from this blog

android - net_scheduler holding wakelock -

sql - MySQL : Getting Entries from a many-to-many table -

java - Retrieving data from database using jsp (Hibernate + Spring + Maven) -