Simulation and Data Analysis – Some interesting patterns
From: [email protected]
Date: 03 Jan 1998
Remote Name: d17-232.uoregon.edu
I am Ph.D. student in the dept od decision sciences at theuniversity of Oregon. Over the last year I have been working withthe lane county 911 call center in an attempt to try anddetermine how many call takers are really needed to satisfy theirrequirements. I found that in looking at the occurrence of 911calls I found that there is a near perfect bimodal normalempherical distribution function i.e. there are two means, notone for the time between calls. the means are marginallycorrelated with the day of the week and time of day that thecalls occur but not enough to explain entirely the existence oftwo means. The bimodal normal nature exists and becomes morepronounced if the calls are stratified by priority level. Theimportance of this concerns the use of simulation as a tool toperform what-if analyses by changing (virtually) the internalorganization of a call center. I have been unable to obtain calldata (time of call, priority of call, type of call, duration ofcall) from other locations due to legal or technicalrestrictions. I would like to try and confirm this bimodalpattern and hopefully explain its existance. Has anyone elseobserved these patterns or does any one have any ideas of how Ican obtain data without violating confidentiality restrictionsthat usually exist with this type of data?
Richard Todd LaSalle (541)346-5105 University of Oregon.Lundquist College of Business Dept. of Decision Sciences.