Comparative Analysis of Customer Behaviour Using K-Means Algorithm Over Convolutional Neural Network with Increase Inaccuracy of Prediction

Abstract

To perform an efficient analysis of customer behaviour by comparing K-means and convolutional neural networks and improving the accuracy of prediction. Material and methods: The sample size for this project as 60 (Group 1=30, Group 2=30). For the customer segmentation behaviour it divides into groups with techniques like Manual, Semi-Automatic, and Automatic. In these techniques, the main problem is low accuracy value rate. To modify this, perform K-means algorithm technique. It allows the data to form a group for individual performance. Result: The data was analyzed to show the density and annual income with significant value (p=0.4512) and it gives the better performance with the validated effects of emotional display on the customers. So that we are able to get performance analysis from our customers and get better reviews. Conclusion: K-means clustering showed improvements in accuracy compared with convolutional neural networks.