The purpose of this section is to layout the
project necessities with a specific end goal to keep up a connection between
the client and the developer. The engineer while following the product
life-cycle stages requires programming’s specifications; this archive will go
about as a reason for the design stage by enumerating every one of the requirements.
This part depicts the current system about ‘Customer Churn Prediction’ that
exists and its constraints and how the proposed framework will beat those
restrictions. What’s more, definite examination of the considerable number of requirements;
functional and non-functional requirements for the proposed framework will be classified.
Today Customer information has properties
of huge samples, big measurements and more commotions. The current researches
or systems about the ‘Customer Churn’ still contain the absence of an
arrangement of logical, framework hypothesis and technique and the single
models strategies for customer churn prediction, likewise can’t totally address
application issues. Subsequently, it is essential for hypothetical and
functional commitments to investigate and think about the client agitate
expectation. The current system of customer churn prediction is done manually
rather than automatically on a designed system for it. The manual prediction
takes an unnecessary amount of time to actually predict something. In most
cases customer churn is not predicted at all. It is essential for an automated
system to be designed which will make this process much easier and overcome the
limitation that most research papers and systems consist of.
The proposed system will overcome the
limitation of the previous work done, researches made and built system about
customer churn prediction, in a better way possible. The proposed system is
built for telecom companies; it can also be adjusted for the customer churn
prediction of any other company as desired. In the proposed system, the user
would take customer data sample. Our system will apply algorithm on the data
and train algorithm according to the provided data. The results will be
obtained after the algorithm is tested on the customer data. The features on the
basis of which the customer churn will be predicted can be weekly traffic,
customer revenue, customer complaints, package info and status. The proposed
model will be trained and tested on previous records of data. Its accuracy can
be from 80-90%. The graphical results can also be shown in the proposed system.
The requirement specifications for customer
churn prediction can be described into two, functional requirements and
The accompanying section explains the vital
functional requirements which should be put together by the framework with the
end goal for it to be deemed as profitable. The functionality is stated
alongside its advantages (benefits), risks and technical problems that may
start amid the procedure of usage.