In practice, the number represents the total error between the fit and the actual data values, squared. To see if the number is "significant" or not, you look the chi square number up in a chi-squared distribution table, according to the number ofĬhi Square curves for various degrees of freedom. Normally, one can think of the value as a number on a chi square distribution curve. The errors are independent of each other and normally distributed. The chi square statistic is used to assess the goodness of fit of curves to the data. If you throw a bit of randomness into your example, the results will make more sense to you. What you see here are floating point round-off errors that are all very, very close to 0, the expected chi square value. In this case, since your data describes a straight line, and since straight lines can be fit more or less exactly, there is no error to calculate. Normally, the chi square goodness of fit statistic calculates how closely a curve fitting function produces output that resembles the actual data. IDL> r = LinFit(x, y, CHISQR=chrsqr) & Print, chisqrĪNSWER: The problem here is how you are selecting theĭata for your "simple" example. IDL> r = SvdFit(x, y, 2, CHISQ=chisqr) & Print, chisqr IDL> r = Poly_Fit(x, y, 1, CHISQ=chisqr) & Print, chisqr But, when I look at a simple example, shown below, I get different values for this keyword in the three routines. This keyword in the on-line documentation seems to be the same: Set this keyword to a named variable that will contain the value of the unreduced chi-square goodness-of-fit statistic. The CHISQR keyword in the IDL routines POLY_FIT, SVDFIT, and LINFIT.
QUESTION: I am trying to understand the meaning of
Coyote's Guide to Traditional IDL Graphics.
0 kommentar(er)
