What math level teaches turning data points into equations?

  1. And what is the process called?

    Say you have a bunch of data you've collected plotted on a graph, what level of math teaches you to convert this data into a formula that you can play with?

    I'm in calc 2 right now and we just covered volumes. It's AWESOME! I'm not much of a higher level math person, but this has got me thinking about moving on. But I really don't know what's out there. We are working with such simple equations, but what if I didn't know the equations?

    Say I plotted the dimensions of 1/2 beer bottle, lengthwise about the x-axis, and wanted to integrate it from 0 to whatever the height is. But the problem is I don't know what formula represents the shape of these dots. I mean I could estimate and do it in sections, but how do I get an equation that represents it perfectly?

    Thanks a lot! Sorry if this is basic :)
     
  2. jcsd
  3. arildno

    arildno 12,015
    Science Advisor
    Homework Helper
    Gold Member

    Why do you think that for any graph there must exist a simple formula that represents it?

    What we CAN do is to approximate the graph by piecewise "simple" formulas and sew them together.

    Because polynomials are so nice to work with, it is quite popular to approximate difficult curves by piecewise polynomials.
     
  4. symbolipoint

    symbolipoint 3,172
    Homework Helper
    Education Advisor
    Gold Member

    Fitting data points to equations is called Mathematical Modeling; also curve fitting. You learn this as early as Introductory Algebra, and you can learn much more of it in Intermediate Algebra and PreCalculus.
     
  5. I think more likely that it is taught more in statistics (around second year of university) and numerical analysis (second or third year of university. The equation to fit data to a curve is quite simple if all you care about is a piece wise approximation. If you wish for derivatives to match up then it may be more complex.
     
  6. I don't, nor did I say this. Currently most of the work we do is with simple graphs (x^2, x^3, lnx, e^x, 1/x, x^(1/2), x, stretched, shrunk, or shifted accordingly) so that we may focus on calculus. I'm interested in the process of deriving a formula for much more complex graphs. I've been reading the wiki on diffy q's and think this might be the answer, although it's a bit over my head.

    actually, after some more searching, I think what I'm looking for is interpolation. That also lead me to info on extrapolation which looks fantastic.
     
  7. uart

    uart 2,776
    Science Advisor

    If your goal is just to find a mathematical formula to fit the points purely for the purpose of finding the integral then you would be much better off looking at numerical integration techniques like Simpsons rule or Runge-Kutter. These work directly on the data points (you could say they implicitly interpolate the data points).
     
    Last edited: Feb 23, 2008
  8. arildno

    arildno 12,015
    Science Advisor
    Homework Helper
    Gold Member

    You can continue with looking in on cubic splines. :smile:
     
  9. Basic polynomial (3rd degree and lower) curve-fitting I did in first year in applied maths, but it was very introductory stuff.
     
  10. The CurvFit program has many data-fitting problems solved for ideas about math models. Plus, it does interpolation and/or extrapolation for ones solution. Source code is include if interested.

    For periodic data use a Sine, Damped Sine, or Lorentzian series as a math model.

    For non-monotonic, but not periodic, data I recommend the Lorentzian series.
     
  11. CRGreathouse

    CRGreathouse 3,682
    Science Advisor
    Homework Helper

    I second arildno's suggestion of cubic splines (or, generally, BĂ©zier splines).
     
Know someone interested in this topic? Share a link to this question via email, Google+, Twitter, or Facebook

Have something to add?