Thread: Choosing an appropriate regression model

I have the date below....

age acid xray size grade nodal
[1,] 66 0.48 0 0 0 0
[2,] 68 0.56 0 0 0 0
[3,] 66 0.50 0 0 0 0
[4,] 56 0.52 0 0 0 0
[5,] 58 0.50 0 0 0 0
[6,] 60 0.49 0 0 0 0
[7,] 65 0.46 1 0 0 0
[8,] 60 0.62 1 0 0 0
[9,] 50 0.56 0 0 1 1
[10,] 49 0.55 1 0 0 0
[11,] 61 0.62 0 0 0 0
[12,] 58 0.71 0 0 0 0
[13,] 51 0.65 0 0 0 0
[14,] 67 0.67 1 0 1 1
[15,] 67 0.47 0 0 1 0
[16,] 51 0.49 0 0 0 0
[17,] 56 0.50 0 0 1 0
[18,] 60 0.78 0 0 0 0
[19,] 52 0.83 0 0 0 0
[20,] 56 0.98 0 0 0 0
[21,] 67 0.52 0 0 0 0
[22,] 63 0.75 0 0 0 0
[23,] 59 0.99 0 0 1 1
[24,] 64 1.87 0 0 0 0
[25,] 61 1.36 1 0 0 1
[26,] 56 0.82 0 0 0 1
[27,] 64 0.40 0 1 1 0
[28,] 61 0.50 0 1 0 0
[29,] 64 0.50 0 1 1 0
[30,] 63 0.40 0 1 0 0
[31,] 52 0.55 0 1 1 0
[32,] 66 0.59 0 1 1 0
[33,] 58 0.48 1 1 0 1
[34,] 57 0.51 1 1 1 1
[35,] 65 0.49 0 1 0 1
[36,] 65 0.48 0 1 1 0
[37,] 59 0.63 1 1 1 0
[38,] 61 1.02 0 1 0 0
[39,] 53 0.76 0 1 0 0
[40,] 67 0.95 0 1 0 0
[41,] 53 0.66 0 1 1 0
[42,] 65 0.84 1 1 1 1
[43,] 50 0.81 1 1 1 1
[44,] 60 0.76 1 1 1 1
[45,] 45 0.70 0 1 1 1
[46,] 56 0.78 1 1 1 1
[47,] 46 0.70 0 1 0 1
[48,] 67 0.67 0 1 0 1
[49,] 63 0.82 0 1 0 1
[50,] 57 0.67 0 1 1 1
[51,] 51 0.72 1 1 0 1
[52,] 64 0.89 1 1 0 1
[53,] 68 1.26 1 1 1 1

What i want to try and do is find an appropriate model for this data. What the data refers to is there are 53 patients in a hospital. Then age, acid, xray, size and grade are all variables to determine whether there is "nodal involvement".... which is something to do with prostate cancer.
So i have age,acid,xray,size and grade which are x1,x2,x3,x4 and x5, regressed upon nodal which is "y".
Im going to do my analysis of it in R after, so im guessing i should maybe use the glm function, family binomial because some of the responses are binary?
Then i possibly need to add some interactions to my model?
Im not really sure how to write down how i would come to such a model.

Any ideas?

You should run a multitude models checking serial correlation and heteroskedasticity. You can use the white test to check and correct for heteroskedasticity. If all goes well with the models, you should then compare F-Stats, T-Stats, and R squareds to find the best model.

eeek, dont understand a few of those terms! For now, all im looking for is an additive model.... Im decent enough at analysisng in R once iv found a model, but dont know im simple terms how to arrive at one?

Can you put that data in an excel sheet and upload it?

mas3309project.doc

I cant get it in excel as i havent got microsoft office.... At the bottom of the document attached is the table of the data. The question im trying to do on here is number 2.
Also if you use R.... This code will create the table for you:

> age<-c(66,68,66,56,58,60,65,60,50,49,61,58,51,67,67,51, 56,60,52,56,67,63,59,64,61,56,64,61,64,63,52,66,58 ,57,65,65,59,61,53,67,53,65,50,60,45,56,46,67,63,5 7,51,64,68)
> acid<-c(0.48,.56,0.50,0.52,0.50,0.49,0.46,0.62,0.56,0.55 ,0.62,0.71,0.65,0.67,0.47,0.49,0.50,0.78,0.83,0.98 ,0.52,0.75,0.99,1.87,1.36,0.82,0.40,0.50,0.50,0.40 ,0.55,0.59,0.48,0.51,0.49,0.48,0.63,1.02,0.76,0.95 ,0.66,0.84,0.81,0.76,0.70,0.78,0.70,0.67,0.82,0.67 ,0.72,0.89,1.26)
> xray<-c(0,0,0,0,0,0,1,1,0,1,0,0,0,1,0,0,0,0,0,0,0,0,0,0, 1,0,0,0,0,0,0,0,1,1,0,0,1,0,0,0,0,1,1,1,0,1,0,0,0, 0,1,1,1)
> size<-c(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 1,1,1,1)
> grade<-c(0,0,0,0,0,0,0,0,1,0,0,0,0,1,1,0,1,0,0,0,0,0,1,0, 0,0,1,0,1,0,1,1,0,1,0,1,1,0,0,0,1,1,1,1,1,1,0,0,0, 1,0,0,1)
> nodal<-c(0,0,0,0,0,0,0,0,1,0,0,0,0,1,0,0,0,0,0,0,0,0,1,0, 1,1,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0,1,1,1,1,1,1,1,1, 1,1,1,1)
> cbind(age,acid,xray,size,grade,nodal)
age acid xray size grade nodal