Find suitable equation for given data...

[rahul8590]

Hey every one ,
well i was writing program which inputs "Statistical data for example
x 1 2 3 4 5
y 1 4 9 16 25
and analysing the data my program must suggest a suitable algebraic equations like looking at this one it should tell me the equation ins y = x square
"

Well wat i have done is i have written different cases like
y = x square
y = log x and so on
but i guess its a very bad style of programming . i would be glad if you guys help me out in suggesting ingenious ways in doing it .

note: "well i know u guys will be a little skeptical about it . but honestly its not my school work or any thing similar to that .

[VernonDozier]

Hey every one ,
well i was writing program which inputs "Statistical data for example
x 1 2 3 4 5
y 1 4 9 16 25
and analysing the data my program must suggest a suitable algebraic equations like looking at this one it should tell me the equation ins y = x square
"

Well wat i have done is i have written different cases like
y = x square
y = log x and so on
but i guess its a very bad style of programming . i would be glad if you guys help me out in suggesting ingenious ways in doing it .

note: "well i know u guys will be a little skeptical about it . but honestly its not my school work or any thing similar to that .

We can't suggest much since you haven't shown us your implementation. Also, we would need a better understanding of the requirements and assumptions. Presumably you are supposed to find the 2nd degree polynomial that best fits a bunch of (x, y) coordinates so that the sum of the errors (I assume that is (predicted x - actual x) squared) is minimized, but that's only an assumption. In the example you gave, that error would be 0. But you have to be more clear about what you are looking for. It's a math/statistics/regression problem first, so you have to understand the methodology, then program it, right?

[rahul8590]

well definetly yes .

well its a statistical problem . i just thought if a given set of coordinated are given , i could predict them how much further they can be extended and find a suitable equation which represents the stuff.
well i could nt show the code for the reason is that its very nasty and brain storming to even undestand wat i am doing in it (although its a piece of mess , so i thought it would be prudent not to show it).
well i would be glad if i am let to know what exactly its called , i mean it must have a technical name for it.

[adam1122]

Like it was said above, we would need to know the requirements of your implementation. If this is an advanced class, it will probably have advanced requirements and thus becomes a more difficult problem.

[adam1122]

well definetly yes .

well its a statistical problem . i just thought if a given set of coordinated are given , i could predict them how much further they can be extended and find a suitable equation which represents the stuff.
well i could nt show the code for the reason is that its very nasty and brain storming to even undestand wat i am doing in it (although its a piece of mess , so i thought it would be prudent not to show it).
well i would be glad if i am let to know what exactly its called , i mean it must have a technical name for it.

Well, what is the class? We need more info.

It could fall under things like regression and machine learning but I suspect this class is not at that level.

[VernonDozier]

well definetly yes .

well its a statistical problem . i just thought if a given set of coordinated are given , i could predict them how much further they can be extended and find a suitable equation which represents the stuff.
well i could nt show the code for the reason is that its very nasty and brain storming to even undestand wat i am doing in it (although its a piece of mess , so i thought it would be prudent not to show it).
well i would be glad if i am let to know what exactly its called , i mean it must have a technical name for it.

I imagine that the technical name you are looking for is "method of least squares"

http://en.wikipedia.org/wiki/Least_squares

It's a big topic, part of a larger topic called "regression analysis".

http://en.wikipedia.org/wiki/Regression_analysis

The first step is to define the problem very precisely so you know what you want to do and where you want to go. Don't even touch the computer programming or even the math before you figure out what you want to do.

[rahul8590]

well sorry for that , its just that i was solving math n wondered if i could program that , n honestly at that moment it felt very easy.
i would definitely work on it.

thanks

[adam1122]

I would encourage you to work on it. If this is your kind of thing, it falls into the broader category of machine learning, which has very interesting applications.

[rahul8590]

well actually i landed up on this program when i was working on high performance computing applications. and other programs which can work on parallel machines and finally i thought such massive huge data of coordinated can be parallelized and thus solved . well i would love to hear more about machine learning .