Churn analysis using deep convolutional neural networks and autoencoders

TL;DR

This paper presents a novel approach to customer churn prediction by transforming behavioral data into images and applying deep learning models, achieving notable accuracy and insights into churn reasons.

Contribution

It introduces a method of representing customer behavior as images for deep learning, combining supervised and unsupervised models to improve churn prediction and understanding.

Findings

Achieved an AUC of 0.743 with convolutional neural networks.

Autoencoders reveal actionable insights into churn reasons.

Effective use of image-based data representation for customer behavior.

Abstract

Customer temporal behavioral data was represented as images in order to perform churn prediction by leveraging deep learning architectures prominent in image classification. Supervised learning was performed on labeled data of over 6 million customers using deep convolutional neural networks, which achieved an AUC of 0.743 on the test dataset using no more than 12 temporal features for each customer. Unsupervised learning was conducted using autoencoders to better understand the reasons for customer churn. Images that maximally activate the hidden units of an autoencoder trained with churned customers reveal ample opportunities for action to be taken to prevent churn among strong data, no voice users.