Real-Time Prediction of Delay Distribution in Service Systems using Mixture Density Networks

TL;DR

This paper introduces a method using Mixture Density Networks to predict the distribution of customer wait times in service systems, enabling improved management and communication of delays.

Contribution

It presents a novel approach combining mixture density modeling with neural networks to accurately predict delay distributions in real-time.

Findings

More delay history improves prediction accuracy.

Mixture Density Networks effectively model wait time distributions.

Predictions support SLA compliance and system management.

Abstract

Motivated by interest in providing more efficient services in customer service systems, we use statistical learning methods and delay history information to predict the conditional distribution of the customers' waiting times in queueing systems. From the predicted distributions, descriptive statistics of the system such as the mean, variance and percentiles of the waiting times can be obtained, which can be used for delay announcements, SLA conformance and better system management. We model the conditional distributions by mixtures of Gaussians, parameters of which can be estimated using Mixture Density Networks. The evaluations show that exploiting more delay history information can result in much more accurate predictions under realistic time-varying arrival assumptions.