Regression

With this routine it is possible to execute a fitting of curves for a series of measurements with a maximum of 1023 pairs of values.

You may choose between the following types of regression and may displace or stretch all points in x- or y-direction, if necessary.

• Proportional Regression                    y = b·x
• Linear Regression                            y = a + b·x
• Polynomial Regression of n-th Order  y = a₀ + ... + a_n · xⁿ
• Geometrical Regression                    y = a·x^b
• Exponential Regression                     y = a·b^x
• Logarithmic Regression                    y = a + b · ln(x)

The function term of the approximation curve, the coefficient of determination, the correlation coefficient and the standard deviation are issued together with the diagram. On the next page, these values are issued once more, together with a table of values of the function.

Example:

Curve through the points P₁(1|1), P₂(2|3), P₃(5|6), P₄(7|3).

Polynomial Regression

 y =  − 6,9152542
        + 4,7189266·x
        − 0,43361582·x2

Coeff.of determin.   = 0,98338318
Correlation coeff.    = 0,99165679
Standard deviation = 0,46028731

See also:

Formulas
Setting the graphics