Suggestions on what machine language algorithm to classify what time a user logs in

Question

I am currently trying to do a project in C# in which i would use Machine Learning to learn and predict the correct time a user will log in. The data that will be logged and captured into the database for learning and predicting will be "User Log In Time" as well as "Number of times user log in around that time period"

I have two part to my question. Firstly if i want to do the above feature, do i need to use supervised or unsupervised machine learning? Because i do not have a data set to train my ML algorithm, i feel i should be using unsupervised.

Secondly if i am using unsupervised, what algorithm will be best suited for this job!

Sorry if i am quite unclear in my question as i just started with machine learning and am very unsure myself! Any help will be greatly appreciated

Answer 1

This sounds like a simple (supervised) regression problem since you try to predict a continuous variable (time), isn't it?

You have login timestamps from different users (your baseline data), separate them by user and add a "trend line". A simple linear regression should be good enough, but you may want to give logistic or polynomial regressions a shoot too.