Normal distribution assumption

Click For Summary

Discussion Overview

The discussion revolves around the assumption of normal distribution for modeling service time in a healthcare context, specifically regarding patient service times. Participants explore theoretical justifications for this assumption without relying on empirical data such as histograms or standard deviation thresholds.

Discussion Character

  • Conceptual clarification
  • Debate/contested
  • Technical explanation

Main Points Raised

  • One participant notes that the exponential distribution is typically used for waiting time problems, highlighting that waiting times cannot be negative, which raises questions about the appropriateness of normal distribution in this context.
  • Another participant provides examples from literature where service times, such as ATM service times, are modeled using normal distribution, suggesting that this approach is not uncommon.
  • A further contribution emphasizes that for normal distribution to be valid, the variation in service time should be evenly distributed around the mean without significant outliers, and that negative values should not arise from subtracting standard deviations from the mean.
  • One participant discusses the central limit theorem, suggesting that the sum of independent identically distributed random variables approaches a normal distribution, which could justify using normal distribution for modeling service times as a sum of various independent tasks.
  • Another participant reflects on the use of normal distribution in cases of deterministic processes with scheduled times, contrasting it with Poisson distribution, but expresses difficulty in finding additional theoretical justifications beyond the sum of independent variables.

Areas of Agreement / Disagreement

Participants express differing views on the appropriateness of normal distribution for modeling service time, with some advocating for its use based on theoretical properties while others raise concerns about its applicability, particularly in relation to waiting times and the presence of outliers.

Contextual Notes

The discussion highlights limitations regarding the assumptions necessary for normal distribution to be valid, such as the need for independence of tasks and the absence of negative values in service time. There is also a noted dependence on the specific characteristics of the service time distribution.

luckyluke
Messages
3
Reaction score
0
Hi everybody.
For sake of work I have to make an assumption that service time for patients is normally distributed.
Normally the doctor has to wait one patient every half hour but anyway this time can vary being less or more.
My teacher wants some theory arguments (normal distribution properties)why we can use normal distribution for modeling service time in our case.
This has to be in generally judging about process without looking on 3sigma,histogram etc.
I don't really know what to say
Any advices about readings or links?
THX
 
Physics news on Phys.org
well the exponential distribution is usually used for waiting time problems, not the normal

unlike the normal, the waiting time cannot be negative
 
Thank you.
It is about service time and there are many examples online where service time follow normal distribution...
For example
cob.jmu.edu/wangpx/kj/MS/5Sim/LectureSimHandout.pdf

where ATM service times follow normal distribution.
or
cs.mwsu.edu/~ranette/CMPS4223-Sim/Statistical%20Distr.ppt
where bus interarrival times are normally distributed.

Regards
 
service time, waiting time have the same issues. just show that the variation in service time is more or less evenly distributed around the mean, without large outlier and the normal will work. Ideally subtracting 2-3 standard deviations from the mean does not get you a negative number. I.e. if the average service time is 1 minute but the SD is 10 minutes due to some large positive outliers then normal will not work. But if its 10 minutes with a standard deviation of 2 minutes normal is OK
 
luckyluke said:
My teacher wants some theory arguments (normal distribution properties)why we can use normal distribution for modeling service time in our case.

A sum of identically distributed independent random variables is approximately normally distributed. I think there are some results relating the sum of independent non-identically distributed random variables to a normal distribution, but I can't quote them. If you think of the service of a patient as involving a sequence of many independent tasks (e.g. weighing the patient involves, walking to the scale, stepping on the scale, reading the scale, recording the weight) you could argue that a normal distribution approximates the sum of the elapsed times of these tasks.
 
Thank you for this advice.
I was reasoning on normal distribution used when there is mistake (late or earlier) from scheduled time.In literature I saw that we can use normal distribution rather than Poisson when we have scheduled time (deterministic process) and this time arrival variate.
Anyway still not finding any other explanations except the fact u mentioned about sum of i.i.d variables.
Regards
 

Similar threads

  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 12 ·
Replies
12
Views
5K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 13 ·
Replies
13
Views
3K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 27 ·
Replies
27
Views
4K