Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I have two equations with this format: $$Ds= A+A^2+\alpha_1\tag{1}$$ and $$Ds= M+M^2+\alpha_2 \tag{2}$$

Knowing that $(1)$ explains 72% of $Ds$ and $(2)$ 20%. I want to combine these two equations into one and know how much this equation explains. Something like:

$$Ds= A+A^2+\alpha_1+M+M^2+\alpha_2$$ (I know it cannot be a sum, but I don't know how to combine this).

Thank you for your answers.

SOSA

P.D: I don't have good notion in mathematics, and I'm not sure about the tag for this question.

Sorry for my bad explanation.

The facts are: I have two factors Age (A) and matrilineal link (M) I have a parameter the David score (Ds, hierarchy rank) I made a quadratic regression with the factor A, and I found that this regression explains 72% of Ds. I did the same thing with the factor M and I found that this regression explains 20% of Ds. So now I want to combine these two equations to explain the parameter Ds with these two factors.

Thank you again for your help

I have this idea: Ds=[(Ds= A+A^2+alpha1) + (Ds= M+M^2+alpha2)]/2

Is it correct? But How can I say how much they explain Ds?

SOSA

share|improve this question
    
What do you mean when you say that an equation explains x% of Ds? –  mkl Dec 18 '12 at 11:44
    
Sorry for my bad explanation. The facts are: I have two factors Age (A) and matrilineal link (M) I have a parameter the David score (Ds, hierarchy rank) I made a quadratic regression with the factor A, and I found that this regression explains 72% of Ds. I did the same thing with the factor M and I found that this regression explains 20% of Ds. So now I want to combine these two equations to explain the parameter Ds with these two factors. Thank you again for your help SOSA –  Sosa Sebastian Dec 18 '12 at 14:14

1 Answer 1

I do not think you have enough information at this point. The reason is the you want to fit the surface

$$ Ds(A,M) = \begin{pmatrix}1\\A\\A^2\end{pmatrix}^\top \begin{vmatrix} K_{11} & K_{12} & K_{13} \\ K_{21} & K_{22} & K_{23} \\ K_{31} & K_{32} & K_{33} \end{vmatrix} \begin{pmatrix}1\\M\\M^2\end{pmatrix}$$

$$ Ds(A,M) = K_{11} + K_{12} M + K_{13} M^2 + K_{21} A + K_{22} A M + K_{23} A M^2 + K_{31} A^2 + K_{32} A^2 M + K_{33} A^2 M^2 $$

and you have only sampled a constant $M=M_\star$ varying $A$ giving you

$$Ds = (K_{11}+K_{12}M_\star+K_{13} M_\star^2) \\ + (K_{21}+K_{22}M_\star+K_{23} M_\star^2) A \\ + (K_{31}+K_{32}M_\star+K_{33} M_\star^2) A^2 \\ = \alpha_1 + \beta_1 A + \gamma_1 A^2$$

and a constant $A=A_\star$ varying $M$ giving you

$$Ds = (K_{11}+K_{21}A_\star+K_{31} A_\star^2) \\ + (K_{12}+K_{22}A_\star+K_{32} A_\star^2) M \\ + (K_{13}+K_{23}A_\star+K_{33} A_\star^2) M^2 \\ = \alpha_2 + \beta_2 M + \gamma_2 M^2$$

Equating your regression coefficients $\alpha_1$, $\beta_1$, $\gamma_1$ and $\alpha_2$, $\beta_2$, $\gamma_2$ to the coefficients $K_{ij}$ of the 3x3 matrix you have 6 equations and 9 unknowns.

You can assume some of the coefficients are zero like $K_{32}=K_{23}=K_{33}=0$ (by ignoring the cross effects of $A$ and $M$) to give you.

$$Ds = (K_{11}+K_{12}M_\star+K_{13} M_\star^2) + (K_{21}+K_{22}M_\star) A + (K_{31}) A^2 \\ = \alpha_1 + \beta_1 A + \gamma_1 A^2$$

and

$$Ds = (K_{11}+K_{21}A_\star+K_{31} A_\star^2) + (K_{12}+K_{22}A_\star) M + (K_{13}) M^2 \\ = \alpha_2 + \beta_2 M + \gamma_2 M^2$$

share|improve this answer
    
Thank you ja72 To be honnest, there is a lot of things that I don't understand (I don't have much knowledge in mathematics). But I have 3 mounths to learn, understand this kind of algebric problem and resolve it. I would like to know, if you don't mind, what is missing to have enought information and resolve the equation ? And, can you tell me in which area of mathématics do I have to look to learn about this kind of problem and resolve it? Thank you very much for your help, I hope you will help me again. Cordially SOSA –  Sosa Sebastian Dec 18 '12 at 19:26
    
Why not simplify the process and go download Eurequa enter your data and it will fit whatever function you want to the data. –  ja72 Dec 19 '12 at 1:45
    
thank you for this information I didn't know that program so I will investigate to see if it solve my problem. Thank you very much –  Sosa Sebastian Dec 19 '12 at 9:19
    
Hi, I checked Eureqa, it seems to be a really good thing to solve my problem. But I will need help for the two steps (prepare the data and set target), I really don't know what to do and why doing it. Can I ask help here? Or do I need to find help in a forum for this software? Thank you very much –  Sosa Sebastian Dec 19 '12 at 14:56
    
@SosaSebastian, I am not sure which forum is appropriate for Eurequa. The comment section for this answer is definitely not the place. Start with the online help and examples for Eurequa. –  ja72 Dec 19 '12 at 16:51

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.