Curve fitting - problem

I have 10000 data sets of (x,y) for which i need to fit a curve. Once I could fit a curve, i need to get points on the curve with an uniform interval. Then I would proceed to patch..and then to stl for my ultimate objective. Right now i'm struck with the curve fitting. Kindly help me out.

Link for the description: https://docs.google.com/open?id=0BxIgxUb0m51-ZDA2ZWU0NzItM2JkZS00MWI4LTg0ZTMtNTI0ZjQzMzIxMzU3

[Walter Roberson]

There are an infinite number of ways to construct a curve through 10000 point pairs. You will not be able to get a meaningful curve fit unless you can put restrictions on the acceptable formula for the curve.

[Navneet Viswan]

Thanks Walter, really makes sense. I couldn`t figure out on how to apply a restriction for this irregular kinda profile. Any suggestions?