Genetic Algorithms

[Ruler2112]

Has anybody here worked with them? I've got a problem that I think needs one and have an idea as to what it's got to do, but haven't a clue as to where to start. (Reading online is what I've been doing so far.)

[qball]

Has anybody here worked with them?

Genetics? or Algorithms?

I've got a problem that I think needs one and have an idea as to what it's got to do

what's problem? how does "Genetic Algorithms" solve, or got to do?

[Ruler2112]

I was wondering if anybody's worked with genetic algorithms. (GA's are a type of AI.)

The problem is that I've got to optimize what options are ordered on vehicles that are to be ordered. To run though and evaluate all possible option combinations would take an exceedingly long time (on the order of centuries), by which time of course the car market will have changed. GAs are able to solve extremely complex problems like this quickly because they evolve generation after generation (or 'learn').

[qball]

GA's are a type of AI

can you give an example of:

GA
AI

????

sandwich/toaster???

The problem is that I've got to optimize what options are ordered on vehicles that are to be ordered.

oops, my bad.

is your problem:

optimize, options, vehicles, orders, or 'to be ordered.'?

[Ruler2112]

AI is Artificial Intelligence- programming a computer so it works more like the human brain than it does, well, a computer. Here's the classic GA problem- a travelling salesman has to visit each of x number of cities and wants to travel the shortest path. To check every path, you'd need to compute and evaluate x! distances. As the number of cities increase, the number of possible permutations becomes exceedingly large- hence a problem that can't be solved with a standard computer program (an NP hard problem). A GA can solve the problem rather quickly.

I need to optimize what options should be ordered on different model vehicles in order for us to produce the most desirable, and therefore quickest to sell, vehicle. Since there are n options available on each model, there are n! option combinations. (It would take on the order of centuries to complete; there are approx. 370,000,000,000,000,000,000,000,000,000,000,000,00 0,000 different possible combinations of options on a vehicle with 36 different option codes.)

There are several sites that give introductory information on these if you're interested; I was wondering if anybody here has ever used them and/or could help me in writing one. I've started on it at work, but am really doing little more than fumbling my way.

[qball]

I think I know what AI is, but, still waiting to find BI, biological intelligence...

travelling salesman has to visit each of x number of cities and wants to travel the shortest path. To check every path, you'd need to compute and evaluate x! distances.

your example is very flawed, here is why:

salesman wants to make sales.
travelling to multiple locations, anyone want to minimize, path, time, $$$.

hence a problem that can't be solved with a standard computer program (an NP hard problem).

how do salesman do it?

hence, if problem not solvable, by any means??? how you going to do?

I need to optimize what options should be ordered on different model vehicles in order for us to produce the most desirable, and therefore quickest to sell, vehicle

oh, I got ya, need to know future, so your prepared.

maybe GA:

offer your product.
monitor results.
adjust, product.
monitor results.
adjust, product.

???
AI THAT!

[Ruler2112]

It's now obvious to me that you have no interest in this except how badly you can crap on my thread qball, so here's my final response to you. (I was thinking you were merely slow to understand instead of just plain stupid, but now I know better.) If anybody else has information as to how to design and/or implement a genetic algorithm, PLEASE feel free to respond. I'm looking for help, somebody to bounce ideas around with, or some other form of intelligent conversation. I really didn't want to be a teacher or a parent in this thread.

your example is very flawed, here is why:

salesman wants to make sales.
travelling to multiple locations, anyone want to minimize, path, time, $$$.

OK. Please write me a program that does not use a GA to determine the shortest path when somebody travels to 200 different cities. See how long your computer takes to complete it. Find a better way to solve the problem without using a genetic algorithm. Also, please explain to the Evolutionary Computing community at large how you've discovered a flaw in the classic problem used to demonstrate the usefullness of this entire branch of computer science. (I'm sure quite a few people who spend obscene amounts of money on research and development would be very interested in hearing this.)

how do salesman do it?

By guessing and by intuitive understanding. This is why our brains are superior to computers.

hence, if problem not solvable, by any means??? how you going to do?

Please read what I wrote instead of what you think I wrote. I wrote that it is not solvable by a standard computer program. (I added bold and italics to point out the key word for you. Please take note.)

oh, I got ya, need to know future, so your prepared.

Not at all. It would help immensely, but what I want to do is to maximize future sales by analyzing past sales.

maybe GA:

offer your product.
monitor results.
adjust, product.
monitor results.
adjust, product.

Yeah, this would be very efficient. Order 200 units costing between $15,000 and $50,000 each, then see how they sell. If they don't sell, no big deal, we'll just change what we order based on no real objective or reproducable method. If this doesn't work, we'll do it again until we're out of money. GREAT METHODOLOGY! The reason I'm writing this program is so that we'll have a measurable, reproducable, objective, consistent, and efficient method of ordering our inventory. The above method accomplishes none of these criteria.

???
AI THAT!

Whatever dude.

[qball]

If anybody else has information as to how to design and/or implement a genetic algorithm, PLEASE feel free to respond

Because you obviously don't!

I was thinking you were merely slow to understand instead of just plain stupid, but now I know better.

So use your newfound knowledge, and be free.

OK. Please write me a program that does not use a GA to determine the shortest path when somebody travels to 200 different cities.

Why would I do that, when, if needed, can just do myself?

Ok, please write me a program that can get somone to 200 cities, by any path?

what's a standard computer program?

is that the one, where you can't find remote and have to change channels on tv by the little buttons?

but what I want to do is to maximize future sales by analyzing past sales.

you need some new technology to do this? what do you think merchants, of all kinds have been doing for 100's of years?

as your text:

so here's my final response to you

good, go do it yourself, 200 cities and all...

[final]
bet you post back
[/final]

[Creosote]

qball--Just give it a rest man. I don't know exactly wtf he is talking about either, but that doesn't mean it is invalid because I have never done what he asks.

Ruler--There might be some other places to ask these questions, such as forums for IT professionals, such as www.Tek-Tips.com or www.Experts-Exchange.com

[Ruler2112]

Thanks for the links Creosote; I'll check them out.