An Effective Customer Behavior Prediction System in Digital Marketing using Ensemble Serial Cascaded Network

Background: Consumer behavior predictions are usually built on the analysis of consumer data, with everyone intended to address a single topic at a time. It is a challenging and unpredictable undertaking to predict client conduct. Methods Used: In this work, a novel customer behavior prediction system is introduced with the support of a deep learning model. Firstly, from the standard datasets, the necessary data are aggregated. Further, the garnered data is fed into the Ensemble Serial Cascaded Network (ESCNet) for predicting customer behaviors. Here, the ESCNet is comprised of an autoencoder, Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM). Results: From the result analysis, the sensitivity of the proposed ESCNet approach is 2.93%, 3.36%, 4.31%, and 3.49% more than 1DCNN, Autoencoder, CNN, and LSTM. Finally, an empirical analysis is acquired in the developed system and compared with the other related techniques to prove the reliability of the designed approach.