XY Scatter

[awibroe]

I a trying to get to grips with the graphing capability of Mathcad.

In the attached I have used a polyfit which obviously is not what I want (I am not sure how to use other graphing functions).

How can I get a line of best fit approach?

[awibroe]

a

[LucMeekes]

Instead of:

Try these:

Then:

Success!

Luc

[Werner_E · Accepted solution]

Ah! I have not read the text thoroughly enough and did not read that a regression line is searched for.

You can also use "polyfit" for that case:

And in case you need the coefficients, you may use "polfitc":

or the built-in "line" command:

[awibroe]

Werner,

Sorry I am really new to this area of Mathcad. In this context what are the coefficients returned?

Andy.

[Werner_E]

> In this context what are the coefficients returned?

The very same as the a and b in Lucs answer.

It was only after reading Lucs correct reply that I realized, that you want to plot a straight regression line and not a more complicated one or a perfect fit (graph runs through

all points) or interpolation.

Luc has shown one way to get the slope a and intercept b with the vertical axis of that line so you can formulate its equation y(x)= a*x+b. He was using the two separate commands "slope" and "intercept" and doing so perfectly documents the role of those two values.

I just showed another way to get those two values with a single command ("line") and also pointed out, that "polyfit" also is able to give you the straight regression line if  you choose polynomial order 1 (highlighted in yellow).

Normally you would not care about the coefficients polyfit uses for its polynomial but you can get Mathcad to tell you those coefficients and a lot of other information using the function "polyfitc" (look it up in the help if you are interested in more details). "polyfitc" returns a matrix with all its information and I simply picked out the two values (slope and intercept) using the command "submatrix". Just another (more complicated) way to get those values.

[Werner_E]

Do you need an exact fit?

Or would a a regression be good enough. Then try order 1, 3, or 5 instead of 8 if you want a polynomial function.

You may also consider an interpolation using splines.

[ValeryOchkov]

See please Study 1. Properties of working fluids, coolants and structural materials for thermal engineering calculations