1

1. ABSTRACT
Customer churn, also known as Customer attrition, customer turnover, or customer defection, is when an existing customer, user, player, subscriber or any kind of return client stops doing business or ends the relationship with a company. Churn prediction is essential for businesses as it helps you detect customers who are likely to cancel a subscription, product or service. To predict whether a customer will be a churner or non-churner, there are several techniques applied for churn prediction, such as artificial neural networks, decision trees, and support vector machines. The past studies in this area indicate that Artificial Neural Networks (ANN) gives more accurate results to predict the customer churn. Therefore, this study will use the ANN to predict customer churn. The data will be collected from the telecom companies and ANN will used to predict the customer churn.
2. INTRODUCTION
The project contains of two main ideas and one particular connecting link between them. We will explain the ideas in simple way then progressively we will get to the details.
The first idea is the customer churn. The customer churn definition is any regular customer to company that depends on the regularity have stopped or cancelled the business or service with them. the term of churn has a lot of meaning but they are all agreed with when your customers or your subscribers stop, canceling their business or services with your product and also for customer loyalty it is an enemy and the sickness that will kill the business. It is calculated by number and the list of customers who left your company, also known as attrition. Why companies must take care of churn to have an idea with the customer who is with your side and to decrease the churn.
Second idea is the neural network the neural network is a computer system influenced by brain nervous.
After defining those two main ideas and accomplish them we must work on the very specific part that represent the main concept of the project which is the prediction. At the prediction we do the most of the art as we go from the neural network into the customer churn.
“The first artificial neuron was produced in 1943 by Warren McCulloch and the logician Walter Pits the neurophysiologist. But one of the most difficult obstacles that affected their progress was the lack of advanced technology, which made it difficult to develop and the first multilayered network was developed in 1975.”(LJUNGEHED, 2017)
“The Neural Networks becomes one of the most development mechanisms that companies interested in development, and this filed started before the arrives of the computers and has standing until these days.”(LJUNGEHED, 2017)

(McDonald, 2017)
3. PROBLEM STATEMENT
The problem is that the companies don’t have a good model to predict the churn of customer and other most don’t know how to predict customer who will churn. Customer retention is generally more cost.
4. ARGUMENTS
The important of the study:
Customer churn impedes growth of a company, so companies should have a well-defined method for calculating customer churn in a given period of time. The prediction can help companies to make strategy for retention of those customer who are going to churn out. And the need of customer lifetime value.
5. AIMS AND OBJECTIVES
Aims:
This research aims to develop a neural network (NN)based customer churn prediction model. Churn prediction model are developed by academics and practitioners to effectively manage and control customer churn in order to retain existing customers by a Neural Network Based Churn Prediction Model.
Objectives:
• Gathering customer information.
• Analysis customer information.
• Insert the requirement into Neural Network system.
• Train the Neural Network system by inserting date the had been gathered.

6. METHODS
• Get the data set of the telecom company.
• Use this data set to train Neural Network.
• Validate the trained Neural Network.

8. TEAMWORK, contribution of each members
The teamwork consists of three members, the team work effectively and efficiency, where that we share tasks fairly, share research and opinions, correct mistakes for each member and discuss complex things.
For example, when we work on literature analysis we shared the work in terms of each member bringing two sources relevant in the chosen topic, also we did the same thing in specifying the objectives and the methods.
9. REQUIREMENTS
• Hardware Not Applicable.
Functional Requirements
• Predicting Customer Churn Using Neural Network technique.
Non-Functional Requirements
Usability: The system should be easy to use.
Reliability: The reliability has unknown percentage to calculate.
Accuracy: The Neural Network system is a system that depends on accurate results.
Performance: The calculation time sand response time should be as little as possible, because the calculation and response time is one of the software features.