Prediction of bank telephone marketing results based on improved whale algorithms optimizing S-Kohonen network

Oct 16, 2026

Uploaded by Noah

Time deposit is characterized by strong stability and low cost, making it a reliable funding source for banks. The paper predicts the success rate of fixed-deposit telemarketing using an S_Kohonen network, formed by adding an output layer to a competition layer in an unsupervised Kohonen model to enable supervised learning. To address instability and limited accuracy from random initialization, an improved whale optimization algorithm optimizes weights between the input and competition layers. Results show higher classification accuracy than GA, WOA, and LWOA-optimized S_Kohonen networks.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Introduction

Time deposit and telemarketing motivation

Prior data-mining studies

Model background and network design

Kohonen network and supervised extension

Optimization approach

Improved whale optimization algorithm

| Prediction of bank telephone marketing results based on improved whale algorithms optimizing S_Kohonen network
Chun Yan a , Meixuan Li a , Wei Liu b ,∗
a College of Mathematics and System Sciences, Shandong University of Science and Technology, Qingdao 266590, China b College of Computer Science and Engineering, Shandong University of Science and Technology, Qingdao 266590, China | | | | --- | --- | --- | | a r t i c l e i n f o | a b s t r a c t
Time deposit has the characteristics of strong stability and low cost. It is a stable source of funds for banks. In this paper, S_Kohonen network is used to predict the success rate of fixed deposit in bank telephone marketing. Firstly, the output layer is added after the competition layer of unsupervised Kohonen network, which makes Kohonen network become S_Kohonen network with supervised learning. Because the improved S_Kohonen network is similar to other feedforward neural networks, each adjacent layer is connected by weights, and the initial weights are random, which easily leads to the unstable output of the network, and still has the disadvantage of relatively low prediction accuracy. Therefore, an improved whale optimization algorithm (IWOA) is proposed to optimize the weights between the input layer and the competition layer of S_Kohonen network. In this paper, the inertia weight of whale optimization algorithm is introduced into random factor on the basis of non-linear decline, and then the random search pattern of Levy flight is introduced into whale algorithm. Finally, the empirical results show that the improved S_Kohonen network can more intuitively represent the classification results of the network, and the classification accuracy of S_Kohonen network optimized by IWOA is significantly higher than that of S_Kohonen network optimized by GA , WOA and LWOA algorithm.
© 2020 Elsevier B.V. All rights reserved. | | | Article history:
Received 22 October 2019
Received in revised form 12 March 2020 Accepted 16 March 2020
Available online 30 March 2020 | | | | Keywords:
Whale optimization algorithm Kohonen neural network Supervised learning | | | 1. Introduction Whether a bank can operate normally or not, the most basic factor is sufficient deposits. Time deposit has the characteristics of strong stability and low cost, so it is a stable source of funds for banks [1]. Time deposit has always been the main source of funds for banks, and the marketing of time deposit has become the daily work of bankers. Therefore, how to improve the success rate of marketing has become an important problem that banks need to solve urgently. Marketing is a powerful means to improve business. With the development of society and the progress of communication, marketing activities have gradually moved from offline to online, and the communication between people has become more convenient. Telemarketing is a low-cost marketing mode, which is widely welcomed by banks. Communication over the telephone can not only make it easier for customers to understand the relevant information of products, but also the level of business. But through the way of aimless and wide-spread network, although it can broaden the crowd of customers, it also causes alot of waste of resources, and even causes harassment to some ∗ Corresponding author. E-mail address: [liuwei_doctor@yeah.net](liuwei_doctor@yeah.net) (W. Liu). users. Therefore, we need to establish effective ways to find target groups and improve the marketing success rate. Evaluate available information and customer indicators through data mining, focusing on maximizing customer lifetime value so that we can build longer and closer tasks with business needs [2]. Selecting potential buyers from a large number of customers is considered a NP-hard problem [3]. In recent years, many scholars have applied data mining technology to improve telemarketing in order to find target groups and improve the success rate of telemark

FAQs

How does the proposed method use an S_Kohonen network for telemarketing prediction?

It adds an output layer after the competition layer of the unsupervised Kohonen network, converting it into an S_Kohonen network with supervised learning to predict fixed-deposit success rates.

Why is an improved whale optimization algorithm introduced?

Because the S_Kohonen network, similar to feedforward neural networks, can produce unstable outputs due to random initial weights and may suffer from relatively low prediction accuracy.

What do the empirical results indicate about the improved approach?

The IWOA-optimized S_Kohonen network achieves significantly higher classification accuracy than versions optimized by GA, WOA, and LWOA, and provides clearer classification outputs.

Oct 16, 2026

Uploaded by Noah