Advertisement
I have a spreadsheet set up with a series of numbers
spreadsheets.google.com/ccc
This represents the number of times I've observed (over 64 choices) someone choosing a Large/Yellow or a Medium/White and so on.
If I use a brute force, I can kinda guess the outcome if there were 36 people or 100.
I do this by dividing the real individual choices by the total of all choices and then multiple that by a target. Using the round function I get whole numbers
=round(((B3/$G$5)*$F$10),0)
But you can see that I don't get exact number
Any Ideas. I wanna do this in Excel, but a description of the real math behind it would be wonderful.
spreadsheets.google.com/ccc
This represents the number of times I've observed (over 64 choices) someone choosing a Large/Yellow or a Medium/White and so on.
If I use a brute force, I can kinda guess the outcome if there were 36 people or 100.
I do this by dividing the real individual choices by the total of all choices and then multiple that by a target. Using the round function I get whole numbers
=round(((B3/$G$5)*$F$10),0)
But you can see that I don't get exact number
Any Ideas. I wanna do this in Excel, but a description of the real math behind it would be wonderful.
posted by:
|
Washington
|
Advertisement
Advertisement
-
Re: Guessing with Brute Force
Sun, February 15, 2009 - 9:00 AMWhat is the choice people are given? Without the beginning of your "problem" your question is unintelligible. -
-
Re: Guessing with Brute Force
Tue, February 17, 2009 - 9:35 AMhi Troy,
i agree, a more complete logical statement about the combinatorics is required.
btw... i appreciate the use of spreadsheeting to create monte carlo studies... very good topic, thank you
-
Re: Guessing with Brute Force
Wed, February 18, 2009 - 6:11 AMThe choices are White, Red, Blue, Yellow, or Green. Each has a size of Small, Medium or Large.
One choice per person:
Person 1 - Blue Medium
Person 2 - Green Small
Person 3 - Blue Small
Person 4 - Red Large
Person 5 - Green Small
-
-
Re: Guessing with Brute Force
Wed, February 18, 2009 - 10:18 AMSo you have a total of 15 possible choices, so a uniform distribution would result in a 1/15 odds for anyone of them to be chosen.
When people choose, they will have a preference for certain combinations, as such, all one can do is observe and report the results. If one changes the set of people who choose, the results could be very different, so there is no exact result. -
-
Re: Guessing with Brute Force
Wed, February 18, 2009 - 10:40 AMWell, usually in statistics you want to find out how different from random the selections are. If it consistently falls into the same pattern DIFFERENT from random, it says something about what you are looking at (for example, if there is something in the picking population that makes them averse to red, it would be chosen less). So the idea is to find some measurements that will indicate how likely the selections made are. For this discrete choice thing, its a little different from continuous values. I'll have to look. I have a great book for this stuff. -
-
Re: Guessing with Brute Force
Tue, February 24, 2009 - 1:38 PMexpectations and confidence
expected values indicate the theoretical probability that a particular combination will be chosen from those which exist.
confidence values the number of data samples which are applied to an experiment to confirm the model.
large data samples cut to include only qualified candidates may offer only a few events, which can limit confidence in measurements of conservation.
neurons used for better rates of progress in analysis of very large samples must match confidence with expectation by application of random numbers in Monte Carlo. Consequently many of the psuedo-random generator code fragments are often insufficient, so procedures and methods for inputs to test neurons are on topic in several good texts. -
-
Re: Guessing with Brute Force
Thu, February 26, 2009 - 5:54 AMI'd like to read more about this model.
Do you have some fav text? -
-
Re: Guessing with Brute Force
Thu, February 26, 2009 - 10:30 AMRandom Number Generation and Monte Carlo Methods (Statistics and Computing)
by James E. Gentle
www.amazon.com/Random-Gen.../ref=sr_1_1
If you are interested in writing software for random number generation and analysis of models, this text may be appropriate. Chapter 8 looks interesting in particular.
If you are asking about combinatorics in general, you would do better to ask someone else.
~Jon
-
-
-
-
-
-
