How do you estimate the best fit of curve?
You always use some model (i.e. function) to fit experimental curve. If you do not know the kind of curve (linear, parabola, Gauss, etc.) you can try to fit with different functions and then compare the residual sum of squares and coefficient of determination of those fit functions. I use MagicPlot for curve fitting, you can try to find something in MagicPlot on-line help.
When should I draw a smooth curve a best fit line or only join the plots (I am especially thinking about Biology for that last one)?
The French curve, or designer's curve, is used for creating garment patterns. Patterns are usually based on standardized sizes intended to fit what's considered to be average sizes. Standardized sizing don't fit everyone, which is why a French Curve is handy. The French Curve can be used to customize garment patterns, allowing sewers to adjust them to fit curvaceous figures or lower a neckline. Examples of uses of a French Curve: Fitting the hips in…
A visual inspection of the scatterplot should tell you whether a straight line or a simple curve is appropriate. You could carry out an analysis of variance (ANOVA) and test the reduction in the residual sums of squares to see whether the curve is significantly better than the line. However, remember that with n observations, a curve with n-1 independent parameters (for example a polynomial of order n-1) will give a prefect fit.
I am calibrating a type S thermocouple from 100 degree C to 800 degree C how many test point do you need to get a good curve fit can someone share your experiences?
If you're trying to plot the curve y = x2, you can do so by plugging a few small values in as x, finding their corresponding y values, and marking those points on a graph. Then draw a "curve of best fit" that intersects all of those points. This particular curve would be a parabola moving upwards, with a focal point of (0, 0).