What section of statistics solves the problem of identification?

Question

Python

Good afternoon, I am engaged in the task of identifying a person as a thesis.

Task: to identify a user (anonymous, authorized) on the site
Introductory: I collect information about the user, such as: geo position, OS version, platform, GMT, hash sum of the canvas area, login if the user is authorized, etag, facebook id
Accordingly, some parameters can change, some parameters tell us exactly what kind of user came to us with a 100% guarantee.
Question: I have a database of already authenticated users with authentication parameters. How can I figure out whether a new user has entered the site, whom we have never identified yet, or a user who visited the site?

What area of ​​statistics will help me solve the problem and tell me which libraries to use in Python?

Answer 1

Of course, statistics do not deal with this. The solution to your problem is a typical classification (diagnosing) problem from the field of Machine Learning. And there "according to metsu" - it is necessary to select the most adequate approach - the Nearest Neighbors Method, Decision Trees, neural networks, etc. - as long as there is enough strength, resources and enthusiasm.
Python libraries - Sklearn, Scipy, first of all. Get into the neural network - Tensorflow, Keras, Pytorch

Answer 2

simple (and primitive): fingerprint + bloom filter