MPD78
03-12-2009, 08:17 AM
Hello all,
In writing an old Fortran program in C++ and actually again in VBA, I came across a set of 3rd and 4th order polynomial equations that were created in the mid 70's through linear regression. Since the properties of the gas are for the most part linear I wondered why I did not find a simple straight line equation until I was informed that the data points were graphed and that the polynomial "looked" the best at the time. Before believing such a poor statement I made a comparison to published values and found that the polynomials were off by an average of 10%. This difference actually is not all that bad because once the properties are used in thier respective correlations the error drops to an average of 2-3% which is the main reason they got away with it.
Anyways, I just wanted to share that with all of you NON-Chi-by-Eye analysts out there.
Thanks
Matt
In writing an old Fortran program in C++ and actually again in VBA, I came across a set of 3rd and 4th order polynomial equations that were created in the mid 70's through linear regression. Since the properties of the gas are for the most part linear I wondered why I did not find a simple straight line equation until I was informed that the data points were graphed and that the polynomial "looked" the best at the time. Before believing such a poor statement I made a comparison to published values and found that the polynomials were off by an average of 10%. This difference actually is not all that bad because once the properties are used in thier respective correlations the error drops to an average of 2-3% which is the main reason they got away with it.
Anyways, I just wanted to share that with all of you NON-Chi-by-Eye analysts out there.
Thanks
Matt