A Secured Predictive Analytics Using Genetic Algorithm and Evolution Strategies

A Secured Predictive Analytics Using Genetic Algorithm and Evolution Strategies

Addepalli V. N. Krishna (Christ University, India), Shriansh Pandey (Christ University, India) and Raghav Sarda (Christ University, India)
DOI: 10.4018/978-1-7998-1290-6.ch009
OnDemand PDF Download:
No Current Special Offers


In the banking sector, the major challenge will be retaining customers. Different banks will be offering various schemes to attract new customers and retain existing customers. The details about the customers will be provided by various features like account number, credit score, balance, credit card usage, salary deposited, and so on. Thus, in this work an attempt is made to identify the churning rate of the possible customers leaving the organization by using genetic algorithm. The outcome of the work may be used by the banks to take measures to reduce churning rates of the possible customers in leaving the respective bank. Modern cyber security attacks have surely played with the effects of the users. Cryptography is one such technique to create certainty, authentication, integrity, availability, confidentiality, and identification of user data can be maintained and security and privacy of data can be provided to the user. The detailed study on identity-based encryption removes the need for certificates.
Chapter Preview


In today’s new age world it is seen that almost everyone now has a bank account with him or her. It is the banks that a provides with the safety of the individuals and their various business assets and they also support with productive human endeavor and economic process by expeditiously and effectively allocating funds, and that they bridge the divergent maturity desires of short-run depositors and long borrowers.

In the last few years it has been seen that in any type of market regaining back the customers are now the major areas of concern. Especially now for the banks due to the increase in the competition in the market due to the entry of new banks that are offering the competing services. It is also a proven fact that a banking client becomes profitable to a bank solely in the second year of his association with the bank and any client that would leave before the period of two years would be a loss making client to the company. If the bank does not offer the customers a powerful reason to remain with them, then the competition can offer them a powerful reason to leave them.

The work is based on paper Genetic Algorithms for feature selection by Kevin R.Coombes where the class prediction package selects a subset of features and combines them to a fully specified Model that can predict the result of newly derived samples. Predictive Models for Bank churning rates were attempted Using Machine learning algorithms like Decision Tree, Gradient Boosting Algorithm, Random Forest and Artificial Neural Networks which have their own Advantages and Limitations. In this work an attempt is made to apply Genetic algorithm and see for its increased performance in its predictions.

Problem Identification

While looking for a suitable topic, we stumbled upon the idea of creating a predictive analysis model, we had decided upon an ideal way of going about it by using the genetic algorithm. We also decided to use ‘R’ as it provided an efficient to provide us with the outcomes. Since Security is an underlying requirement for any data Model we thought of applying attribute based cryptography (IBE) for security purpose.

Problem Formulation

As mentioned earlier the increasing churn rate is the new age problems for the banks in the current market system. So it is the work of the bank to keep a close eye on the banking customers and should also fulfill their needs on the customers who are more probable to leave the bank. The factors that would determine is by watching out for the following factors like customer name, credit score, tenure, age, balance, number of products has a card or not. The aforementioned factors can be used to find out the number of customers or the banking clients who are going to stay with the bank using the data-set available to us.

The work requires us to use these data-sets in order to find out the banking clients which are going to stay with the bank. Since it is a new way of predictive analysis it is expected to get good results and getting more accuracy than the existing system.

The reason for using the genetic algorithm was as, we have seen that GA totally works on the basis of feature selection means it will select those features (attributes) from the dataset which are the most important for the churning rates of customer and because of that the results will be more precise and accurate. Side by side we have used our own model (logic) that how to make a table of data set, so that GA gets optimized and can produce more good results.

The work is appended with IBE based Security as the top layer, which forms an integral part with any data based Model in increasing its performance.

Problem Statement and Objectives

In the banking industry regaining of customers is the most important thing for a bank so as to gain benefits due to the engagement of the baking clients. In order to know about the problem we have taken a bank data set which has been collected by a bank after doing certain sort of surveys. The bank has been seen the unusual churn rates. They found out that the customers are leaving the bank at very high and unusual rates, so the bank wants to understand that what the particular problem is behind the high churning rates. In order to provide security to the data, attributes to be identified for which we tried with Identity based encryption.

Complete Chapter List

Search this Book: