[R] Stepwise logistic discrimination

[Peter Ho]

Dear All R users,

Firstly , I would like to wish all R users a Merry Xmas and  a happy New Year.

I have been trying to do stepwise logistic discrimination on a data set of 126
variables and only 99 cases, into 4 groups. I followed Prof. Riply's suggestion of
applying stepAIC after using multinom from nnet, but I got an error and hence I
cannot go any further with the analysis. Does anyone out there have any suggestions
as to where I went wrong? Below is a short summary what I did and the results,
ending with the error message.  I am currently using R 0.64.2 (windows binary).


------------------
> nose126s.mu <- multinom(Spoilage ~ ., nose126s)
# weights:  512 (381 variable)
> nose126s.mu
Call:
multinom(formula = Spoilage ~ ., data = nose126s)

Coefficients:
        (Intercept)        FRAG25        FRAG26       FRAG27        FRAG28
week2  1.219487e-09 -3.195595e-04  4.748949e-05 4.900736e-05 -8.570984e-08
week3 -1.660521e-09 -1.669241e-05 -1.557546e-04 1.422205e-04  1.666224e-06
week4 -3.602248e-10  6.105686e-06  2.218175e-04 6.301120e-06  1.859483e-07

....
            FRAG149       FRAG150
week2 -0.0001381260 -2.165186e-05
week3 -0.0001115298 -1.141151e-04
week4 -0.0001030075  1.792253e-05

Residual Deviance: 19.53574
AIC: 613.5357

> nose126s.step <- stepAIC(nose126s.mu)
Start:  AIC= 613.54
 Spoilage ~ FRAG25 + FRAG26 + FRAG27 + FRAG28 + FRAG29 + FRAG30 +
    FRAG31 + FRAG32 + FRAG33 + FRAG34 + FRAG35 + FRAG36 + FRAG37 +
    FRAG38 + FRAG39 + FRAG40 + FRAG41 + FRAG42 + FRAG43 + FRAG44 +
    FRAG45 + FRAG46 + FRAG47 + FRAG48 + FRAG49 + FRAG50 + FRAG51 +
    FRAG52 + FRAG53 + FRAG54 + FRAG55 + FRAG56 + FRAG57 + FRAG58 +
    FRAG59 + FRAG60 + FRAG61 + FRAG62 + FRAG63 + FRAG64 + FRAG65 +
    FRAG66 + FRAG67 + FRAG68 + FRAG69 + FRAG70 + FRAG71 + FRAG72 +
    FRAG73 + FRAG74 + FRAG75 + FRAG76 + FRAG77 + FRAG78 + FRAG79 +
    FRAG80 + FRAG81 + FRAG82 + FRAG83 + FRAG84 + FRAG85 + FRAG86 +
    FRAG87 + FRAG88 + FRAG89 + FRAG90 + FRAG91 + FRAG92 + FRAG93 +
    FRAG94 + FRAG95 + FRAG96 + FRAG97 + FRAG98 + FRAG99 + FRAG100 +
    FRAG101 + FRAG102 + FRAG103 + FRAG104 + FRAG105 + FRAG106 +
    FRAG107 + FRAG108 + FRAG109 + FRAG110 + FRAG111 + FRAG112 +
    FRAG113 + FRAG114 + FRAG115 + FRAG116 + FRAG117 + FRAG118 +
    FRAG119 + FRAG120 + FRAG121 + FRAG122 + FRAG123 + FRAG124 +
    FRAG125 + FRAG126 + FRAG127 + FRAG128 + FRAG129 + FRAG130 +
    FRAG131 + FRAG132 + FRAG133 + FRAG134 + FRAG135 + FRAG136 +
    FRAG137 + FRAG138 + FRAG139 + FRAG140 + FRAG141 + FRAG142 +
    FRAG143 + FRAG144 + FRAG145 + FRAG146 + FRAG147 + FRAG148 +
    FRAG149 + FRAG150

# weights:  508 (378 variable)
# weights:  508 (378 variable)

......


Error: arguments imply differing number of rows: 1, 126

---------------------------


Please can anyone suggest a way forward?


Thank you


Peter
-------------------
Peter Ho
Escola Superior de Biotecnologia
Rua Dr. António Bernardino de Almeida
4200 Porto
Tel: ++351-22-5580043



-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._