Digital twin in consumer choice modeling

Abstract

Listen

Translate article

Abstract “Digital twin” more often perceived as a twin terminology along with industry virtualization of physical assets. The usage of digital twin on physical asset is well known, such as to predict when the individual parts of a machine must be replaced. However, digital twin technology in non-physical modeling is a vibrant research area. One area where digital twin can be effective is predicting the customer's needs. Most businesses to predict the customer's needs uses risk analysis and profitability assessment which holds its own pitfalls. One of the major downfalls arises during the analysis on historical data is the time consumed. Time is one of the crucial factors that determines the profit a company makes, holding of customers, satisfying the customers' needs at the right time, fails because of the static behavior. This can be made more effective by enforcing Digital twin to track the customer behavior dynamically such as the products they consume, their satisfaction. So instead of relying on the historical data, the data for digital twin will be from CRMs, logs, order processing info etc. Right product at right time can be achieved by creating suitable machine learning models on this dynamic dataset and this trained model are held in the digital twin, which runs them in real time. For achieving this approach, a specific technology called Tarantool Data Grid is very useful. In this chapter, we will explore how this technology can be used to create consumer choice modeling using Digital Twin with suitable use cases.