11.08.2013 Views

CALL CENTERS (CENTRES) - Faculty of Industrial Engineering and ...

CALL CENTERS (CENTRES) - Faculty of Industrial Engineering and ...

CALL CENTERS (CENTRES) - Faculty of Industrial Engineering and ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

tions to Hayward’s approximation, generalized peakedness, asymptotic expansions for the Erlang<br />

loss function, the normal-distribution method, <strong>and</strong> bounds for the blocking probability. For the<br />

case <strong>of</strong> no extra waiting space, a renewal arrival process <strong>and</strong> an exponential service-time distribution<br />

(the GI/M/s/O model), a heavy-traffic local limit theorem by A.A. Borovkov implies that<br />

the blocking depends on the arrival process only through the first two moments <strong>of</strong> the renewal<br />

interval as the <strong>of</strong>fered load increases. Moreover, in this situation, Hayward’s approximation is<br />

asymptotically correct.<br />

Keywords: Probability, Queueing theory, Telecommunication traffic, Service systems, Blocking,<br />

Blocking probability, Congestion measures, Servers, Waiting spaces, G/GI/s/r model, Heavy<br />

traffic limit theorems, Conditioning heuristic, Hayward’s approximation, Peakedness, Asymptotic<br />

expansions, Erlang loss function, Normal distribution method, Exponential service time<br />

distribution, GI/M/s/O model, Arrival process<br />

15. Mabert, V.A. Short interval forecasting <strong>of</strong> emergency phone call (911) work loads, Journal <strong>of</strong><br />

Operations Management, 5 (3), 1985, 259–271.<br />

Abstract. There has been a growing emphasis over the last 5–10 years on improving productivity<br />

in the service sector <strong>of</strong> the US economy. Effective scheduling <strong>of</strong> the workforce in these<br />

organizations requires good estimates <strong>of</strong> dem<strong>and</strong>, which may show substantial variations between<br />

days for certain times <strong>of</strong> the year. An examination is made <strong>of</strong> the use <strong>of</strong> 6 different forecasting<br />

methods for predicting daily emergency call workloads for the Indianapolis Police Department’s<br />

communications area: 1. one-year lag, 2. zero/one regression, 3. multiplicative/additive, 4.<br />

zero/one with adjustment, 5. multiplicative/additive with adjustment, <strong>and</strong> 6. autoregressive,<br />

integrated moving average intervention. The research suggests that there are clearly significant<br />

differences in performance for the 6 models analyzed. Simple modeling approaches can perform<br />

well in the complex environments found in many service organizations. Special tailoring <strong>of</strong> the<br />

forecasting model is required for many service firms. Historical data patterns for these organizations<br />

tend to be more involved than just trend <strong>and</strong> seasonal elements.<br />

Keywords: Studies, Police, Mathematical models, Implementations, Forecasting techniques,<br />

Emergencies, Departments, Communications, Case studies<br />

(Appears also in Section II.)<br />

16. H<strong>of</strong>fman, K.L. <strong>and</strong> C.M. Harris. Estimation <strong>of</strong> a caller retrial rate for a telephone information<br />

system, European Journal <strong>of</strong> Operational Research, 27 (2), 1986, 207–214.<br />

Abstract. As part <strong>of</strong> a continuing study <strong>of</strong> the usage <strong>of</strong> its Taxpayer Service Telephone Network,<br />

the US Internal Revenue Service wished to determine more accurate methods for dem<strong>and</strong><br />

measurement. It has long been recognized that the total number <strong>of</strong> calls coming into such a busy<br />

telephone system overestimates the actual number <strong>of</strong> distinct callers. The service had previously<br />

estimated its real dem<strong>and</strong> by adding ( 1<br />

3 ) <strong>of</strong> both the number <strong>of</strong> blocked or overflow calls <strong>and</strong><br />

the number <strong>of</strong> ab<strong>and</strong>onments to the total actually answered. The thrust <strong>of</strong> this current study<br />

then was to develop an accurate statistical method for providing a more objective formula for<br />

this true dem<strong>and</strong>, which turns out to be equivalent to estimating the probability <strong>of</strong> retrial by<br />

blocked <strong>and</strong> ab<strong>and</strong>oned callers.<br />

6

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!