search

SimonRinguette / bpswg

Automatically exported from code.google.com/p/bpswg
0 stars 0 forks source link

Implementer's guide: comparison with 'real-life'... #118

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
Ref: Sec 1.3
The statement "Similarly comparing a simulation generated set of results 
against ‘real life’ is also potentially misleading" does not seem to take 
into account that 'real-life' data may be accurate, if the data comes from 
running and monitored processes (assuming that is never the case?).

Original issue reported on code.google.com by razvan.r...@wwhc.us on 31 Jan 2013 at 2:42

GoogleCodeExporter commented 9 years ago 
Replaced the mentioned paragraph with the following: 

Whilst the purpose of simulation is typically to improve on an outcome from 
'real-life', care should be taken to ensure that the data used is 
representative. Obviously that real historical data represents a snapshot that 
may be either more optimistic or more pessimistic than is typical. More subtly, 
using real data may either include or exclude rare outliers (data points 
significantly different to typical).  Hence it is theoretically more sound to 
determine the distribution that the sample comes from. This allows you to run 
multiple replications based upon the execution history, as opposed to running 
the execution history which is only a single replication.

Original comment by tstephen...@gmail.com on 2 Mar 2013 at 12:14