Accurate Traffic/ Subscriber Prediction

[justdream]

Dears,

Do you know any tool/ solution to do an Accurate Traffic/ Subscriber Prediction [that you already tested with good prediction]
*Regardless of the vendor [as most probably input will be statistical data]

[coach]

A general approach:

Use the traffic trend per cell on busy hour as an input to a linnear regression model, there you can get an estimation on what can be the expected traffic growth for each cell individually going forward. Note that you should build a linear regression model for each cell, not one for the whole network.

This can be easily done with Excel, or some data-science-friendly programming language like Python or R.

When doing this, its important to take enough data (at least a year, maybe more) and to “clean” the data (for example: if a site was OOS one week, just consider for that week an average value), this is to avoid messing the linnear regression model.

The advantage of this method is that you will get the trend for each prticular cell. It is possible that the traffic growth trend is not the same for all sites (site A may grow 10% in a year, but site B may grow 30%, for example)

[dekili]

Hello,

I agree with the coach that you would need some kind of data science to build a model and make inference. As I am aware, check Cardinality (UK company), their website for potential tool that would fit your need.

dekili