Browse Prior Art Database

IMPUTED AVAILABILITY MODEL

IP.com Disclosure Number: IPCOM000249722D
Publication Date: 2017-Mar-28
Document File: 8 page(s) / 909K

Publishing Venue

The IP.com Prior Art Database

Related People

Xi Gong: AUTHOR [+3]

Abstract

A resource's relative availability is imputed using historical data instead of relying on a human's assertion of availability. This is a statistical model used to aid a decision support system.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 25% of the total text.

Copyright 2017 Cisco Systems, Inc. 1

IMPUTED AVAILABILITY MODEL

AUTHORS: Xi Gong

Ryan LaFountain Tom Willingham

CISCO SYSTEMS, INC.

ABSTRACT

A resource's relative availability is imputed using historical data instead of relying

on a human's assertion of availability. This is a statistical model used to aid a decision

support system.

DETAILED DESCRIPTION

Optimizing resource utilization has always been a goal of customer contact business

operations. However, problems exist with regard to accurately determining resource

availability. This may be caused by human behavior or the inability of current systems to

determine relative, instead of binary, availability.

Many systems that route work to resources rely on some measure of resource

availability, typically allowing the resource to assert availability in a binary way. For

example, a call center agent may tell the system that he or she is ‘available’ or ‘unavailable’

for new work. However, problems arise when this human asserted availability is inaccurate.

For instance, if an agent forgets to tell the system he or she is ‘unavailable,’ customer calls

may continue to be routed to this off shift agent.

Furthermore, when relying solely on asserted availability, resources may be

determined to be equally available/unavailable. In reality, resources working on multiple

pieces of work at once, with pieces of work taking varying times complete, have not binary,

but relative, availability. For example, a call center agent may be working on a greater

number of tasks that take a longer amount of time to complete than a peer agent. A system

with only a basic understanding of binary availability may see these agents as equally

available and route a new piece of work to the agent who is less available.

As such, the techniques described herein impute a resource’s relative availability

using historical data instead of relying on a human’s assertion of availability. They use a

statistical model to aid a decision support system in determining a resource’s availability

Copyright 2017 Cisco Systems, Inc. 2

by determining (1) a confidence factor for a resource’s asserted availability (i.e., how much

should the system trust a resource’s assertion that they are available or unavailable?); and

(2) a comparative probability that a resource is available for a new piece of work (i.e., how

available is a resource compared to that resource’s peers?).

This system uses probability models and k-means clustering techniques to enhance

its understanding of resource availability. Conventionally, a resource must tell the system

he or she is ‘available’ or ‘unavailable’. By contrast, this system may impute the confidence

of the human asserted binary availability and may also predict when a resource is

comparatively available to his or her peers to accept a piece of work. The entire system

may include two core components: (1) Availability Regeneration Function (Yo-yo

function), and (2) Profile Usage Change Reliability (PUCR). The firs...