Random Forest Notes -- for lecture 4/26/24
  ------------------------------------------

> fullmodP$call
glm(formula = Yv ~ ., family = binomial(link = "probit"), 
    data = dfr2)
> round(summary(fullmod)$coef[,1:3], 5)
            Estimate Std. Error   z value
(Intercept) -3.58665    0.17427 -20.58149
X1           1.32353    0.16504   8.01923    ** Mysterious why not important in trees ?
X2           0.50019    0.11517   4.34318
X3           0.99929    0.28118   3.55391
X4           1.34232    0.14382   9.33358    ***
X5           0.93741    0.15820   5.92550
X6           2.36205    0.24631   9.58983    ***
X7          -0.00532    0.31620  -0.01682
X8           0.25574    0.15230   1.67916    ## Not very important by itself ?!
X9           0.53758    0.05622   9.56197
X12         -1.26766    0.09390 -13.50006    ## This interaction did not come out in trees ?!
X13         -1.51108    0.26704  -5.65868
X15         -0.89008    0.18778  -4.73992
X17         -0.74162    0.24136  -3.07272
X18         -0.17270    0.09176  -1.88207
X24         -1.28916    0.14604  -8.82724
X26         -0.98707    0.24867  -3.96943
X27         -0.80831    0.24810  -3.25804
X34          0.22732    0.25256   0.90006
X35         -0.82562    0.29958  -2.75597
X36         -0.50721    0.42691  -1.18809
X38         -1.36637    0.26516  -5.15291
X47         -0.07985    0.39491  -0.20220
X48         -1.23328    0.12913  -9.55072    ## May account for special X4, X8 role in trees
X58         -0.48910    0.15714  -3.11263
X67         -0.97484    0.68593  -1.42120
X68         -0.97191    0.22241  -4.36993
X127        -1.60922    0.30989  -5.19284
X135        -1.50800    0.36257  -4.15913
X138        -1.39847    0.29224  -4.78540    ## Also helps account for importantce of X8
X358         0.18404    0.32845   0.56033
X1358        0.63533    0.34328   1.85076

### SUCCESS OF MACHINE LEARNING METHODS OFTEN DEPENDS ON A VERY LOW DIMENSIONAL
### SUBMODEL DOING MOST OF THE PREDICTIVE WORK !!