How to calculate coefficients for cubic least squares regression utilizing equations for summation of x- and y- data points only.

263 Views Asked by At

For quadratic least squares statistical regression equations, the following is utilized for the calculation of coefficients $(a,b,c)$:

$a = \cfrac{∑(x^2y) * ∑(xx)-∑(xy) * ∑(xx^2)}{∑(xx) * ∑(x^2x^2)-∑(xx^2)^2}$

$b = \cfrac{ ∑(xy) * ∑(x^2x^2) - ∑(x^2y) * ∑(xx^2)}{∑(xx) * ∑(x^2x^2)- ∑(xx^2)^2}$

$c = \text{mean}(y) - b * \text{mean}(x) - a * \text{mean}(x^2)$

where,

$∑(xx) = ∑(x^2) - \cfrac{∑(x)^2}k$

$∑(xy) = ∑(xy) - \cfrac{∑(x)*∑(y)}k$

$∑(xx^2) = ∑(x^3) - \cfrac{∑(x)*∑(x^2)}k$

$∑(x^2y) = ∑(x^2y) - \cfrac{∑(x^2)*∑(y)}k$

$∑(x^2x^2) = ∑(x^4) - \cfrac{∑(x^2)^2}k$

Can anyone help me identify this same method for finding the coefficients $(a,b,c,d)$ with regard to cubic regression statistical equations?

Like this:

$a = ???$

$b = ???$

$c = ???$

$d = \text{mean}(y) - c * \text{mean}(x) - b * \text{mean}(x^2) - a * \text{mean}(x^3)$

where, the same identities as above, but possibly including $∑(x^3y)$, $∑(x^2x^3)$ and $∑(x^3x^3)$?