Species 064: Juniperus occidentalis

General Linear Models

Boosted Regression Trees

Generalized Linear Model Results

Data:

C:\Withrow\Workspace\Model\glm_1\CovariateCorrelationOutputMDS_initial.csv

n(pres)                 =      353

n(abs)                  =      4651

n covariates considered =      7

total time for model fitting = 1.44min

Settings:

model family          : binomial

simplification method : AIC

Results:

number covariates in final model   : 24

Call:

glm(formula = response ~ bio08 + I(bio08^2) + I(bio04^2) + bio02 + 

    bio04 + I(di_all_model^2) + I(bio17^2) + bio17 + I(bio02^2) + 

    bio09 + I(bio09^2) + di_all_model + bio02:bio09 + bio02:di_all_model + 

    bio08:bio04 + bio04:bio09 + bio08:bio09 + bio08:bio02 + bio02:bio17 + 

    bio02:bio04 + bio17:di_all_model + bio09:di_all_model + bio08:di_all_model + 

    bio04:di_all_model, family = model.family, data = dat, weights = weight, 

    na.action = "na.exclude")

Deviance Residuals: 

   Min      1Q  Median      3Q     Max  

-1.652  -0.081   0.000   0.000   3.390  

Coefficients:

                     Estimate Std. Error z value Pr(>|z|)    

(Intercept)        -1.822e+02  2.033e+01  -8.966  < 2e-16 ***

bio08               1.125e+00  3.286e-01   3.424 0.000616 ***

I(bio08^2)         -1.246e-02  2.330e-03  -5.347 8.95e-08 ***

I(bio04^2)         -1.490e-05  2.333e-06  -6.385 1.71e-10 ***

bio02               6.526e-01  2.040e-01   3.198 0.001382 ** 

bio04               7.285e-02  9.651e-03   7.548 4.41e-14 ***

I(di_all_model^2)  -6.561e-04  2.809e-04  -2.335 0.019518 *  

I(bio17^2)         -1.738e-03  5.829e-04  -2.981 0.002868 ** 

bio17              -3.379e-01  1.739e-01  -1.943 0.051995 .  

I(bio02^2)         -2.501e-03  6.348e-04  -3.941 8.13e-05 ***

bio09              -1.202e+00  2.806e-01  -4.285 1.83e-05 ***

I(bio09^2)         -7.639e-03  1.590e-03  -4.806 1.54e-06 ***

di_all_model       -7.464e-01  1.773e-01  -4.211 2.55e-05 ***

bio02:bio09        -5.304e-03  2.252e-03  -2.355 0.018501 *  

bio02:di_all_model  2.854e-03  9.186e-04   3.106 0.001894 ** 

bio08:bio04        -8.305e-04  1.410e-04  -5.889 3.89e-09 ***

bio04:bio09         6.570e-04  1.153e-04   5.697 1.22e-08 ***

bio08:bio09         1.887e-02  3.717e-03   5.077 3.83e-07 ***

bio08:bio02         8.937e-03  2.641e-03   3.383 0.000716 ***

bio02:bio17         2.884e-03  9.989e-04   2.887 0.003885 ** 

bio02:bio04         1.310e-04  6.716e-05   1.951 0.051103 .  

bio17:di_all_model  1.982e-03  1.059e-03   1.871 0.061278 .  

bio09:di_all_model  5.821e-03  1.886e-03   3.086 0.002028 ** 

bio08:di_all_model -5.373e-03  2.057e-03  -2.613 0.008987 ** 

bio04:di_all_model -1.105e-04  5.140e-05  -2.149 0.031624 *  

---

Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 2552.5  on 5003  degrees of freedom

Residual deviance: 1264.1  on 4979  degrees of freedom

AIC: 1314.1

Number of Fisher Scoring iterations: 22

============================================================ 

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.6081446 

NULL Deviance                : 0.51009

Fit Deviance                 : 0.25261

Explained Deviance           : 0.25747

Percent Deviance Explained   : 50.476

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.14 

Confusion Matrix: 

         observed

predicted    1    0

        1  317  496

        0   36 4155

AUC                          :  0.9555 

Percent Correctly Classified :  89.36851 

Sensitivity                  :  0.898017 

Specificity                  :  0.8933563 

Kappa                        :  0.493958 

True Skill Statistic         :  0.7913733 

   Calibration Statistics 

Intercept (general calibration)                            :  0.0008197111 

Slope   (direction and variation in fit)                   :  1.001387 

Testa0b1 (overall reliability of predictors)               :  0.9997452 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9972167 

Testb1|a (refinement given correct calibration)            :  0.9822035 

============================================================ 

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.5731177  (sd 0.038208) 

NULL Deviance                : 0.51007 (sd 0.0046687)

Fit Deviance                 : 0.27393 (sd 0.025131)

Explained Deviance           : 0.23614 (sd 0.023653)

Percent Deviance Explained   : 46.308 (sd 4.7552)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.146 (sd 0.005164) 

Confusion Matrix: 

         observed

predicted    1    0

        1  306  512

        0   47 4139

AUC                          :  0.9472  (sd 0.01056) 

Percent Correctly Classified :  88.8297  (sd 1.8371) 

Sensitivity                  :  0.8671429  (sd 0.055428) 

Specificity                  :  0.8899215  (sd 0.02) 

Kappa                        :  0.4733678  (sd 0.052685) 

True Skill Statistic         :  0.7570644  (sd 0.055732) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.1033484  (sd 0.24965) 

Slope   (direction and variation in fit)                   :  0.8836934  (sd 0.16397) 

Testa0b1 (overall reliability of predictors)               :  0.5259658  (sd 0.38141) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.6515437  (sd 0.29262) 

Testb1|a (refinement given correct calibration)            :  0.4341081  (sd 0.37906) 

Total time = 8.94 min

Boosted Regression Tree Modeling Results

Data:

C:\Withrow\Workspace\Model\brt_1\CovariateCorrelationOutputMDS_initial.csv

n(pres)                 =      353

n(abs)                  =      4651

n covariates considered =      7

total time for model fitting = 0.31min

Settings:

random seed used             : 19680828

tree complexity              : 9

learning rate                : 0.0587

n(trees)                     : 300

model simplification         : cross-validation

n folds                      : 3

n covariates in final model  : 4

Relative influence of predictors in final model:

          Var  rel.inf

        bio04 33.76958

        bio02 29.30845

        bio08 26.21656

 di_all_model 10.70541

Important interactions in final model:

 v1 name1 v2 name2

  3 bio08  1 bio02

  3 bio08  2 bio04

============================================================ 

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.8727186 

NULL Deviance                : 0.51009

Fit Deviance                 : 0.12113

Explained Deviance           : 0.38896

Percent Deviance Explained   : 76.253

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.18 

Confusion Matrix: 

         observed

predicted    1    0

        1  340  184

        0   13 4467

AUC                          :  0.9938 

Percent Correctly Classified :  96.06315 

Sensitivity                  :  0.9631728 

Specificity                  :  0.9604386 

Kappa                        :  0.7546915 

True Skill Statistic         :  0.9236114 

   Calibration Statistics 

Intercept (general calibration)                            :  0.5276852 

Slope   (direction and variation in fit)                   :  1.845166 

Testa0b1 (overall reliability of predictors)               :  0 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9902529 

Testb1|a (refinement given correct calibration)            :  0 

============================================================ 

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6210477  (sd 0.065534) 

NULL Deviance                : 0.51007 (sd 0.0046687)

Fit Deviance                 : 0.26887 (sd 0.034451)

Explained Deviance           : 0.2412 (sd 0.033205)

Percent Deviance Explained   : 47.303 (sd 6.6032)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.209 (sd 0.025144) 

Confusion Matrix: 

         observed

predicted    1    0

        1  256  248

        0   97 4403

AUC                          :  0.95553  (sd 0.010604) 

Percent Correctly Classified :  93.10567  (sd 1.1192) 

Sensitivity                  :  0.7254762  (sd 0.070583) 

Specificity                  :  0.9466791  (sd 0.01109) 

Kappa                        :  0.5621225  (sd 0.059811) 

True Skill Statistic         :  0.6721553  (sd 0.070796) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.1311575  (sd 0.23045) 

Slope   (direction and variation in fit)                   :  0.7983736  (sd 0.080029) 

Testa0b1 (overall reliability of predictors)               :  0.1755564  (sd 0.19542) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.5408579  (sd 0.26312) 

Testb1|a (refinement given correct calibration)            :  0.1554098  (sd 0.25594) 

Total time = 3.52 min

MARS

Random Forests

MARS Model Results

Data:

C:\Withrow\Workspace\Model\mars_1\CovariateCorrelationOutputMDS_initial.csv

n(pres)                 =      353

n(abs)                  =      4651

n covariates considered =      7

total time for model fitting = 0.05min

Settings:

random seed used             : 123

mars degree                  : 1

mars penalty                 : 2

Summary of Model:

      nsubsets   gcv    rss

bio04       13 100.0  100.0

bio08       10  59.7   60.8

bio02        9  49.5   51.0

bio17        7  29.2   31.7

bio09        6  26.2   28.6

bio06        5  24.2   26.3

============================================================ 

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.5733663 

NULL Deviance                : 0.51009

Fit Deviance                 : 0.27312

Explained Deviance           : 0.23697

Percent Deviance Explained   : 46.457

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.12 

Confusion Matrix: 

         observed

predicted    1    0

        1  313  538

        0   40 4113

AUC                          :  0.9485 

Percent Correctly Classified :  88.44924 

Sensitivity                  :  0.8866856 

Specificity                  :  0.884326 

Kappa                        :  0.4667576 

True Skill Statistic         :  0.7710115 

   Calibration Statistics 

Intercept (general calibration)                            :  0.000447389 

Slope   (direction and variation in fit)                   :  1.000634 

Testa0b1 (overall reliability of predictors)               :  0.9999164 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9972466 

Testb1|a (refinement given correct calibration)            :  0.9900544 

============================================================ 

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.5715726  (sd 0.033262) 

NULL Deviance                : 0.51007 (sd 0.0046687)

Fit Deviance                 : 0.2765 (sd 0.015927)

Explained Deviance           : 0.23357 (sd 0.015171)

Percent Deviance Explained   : 45.795 (sd 3.0199)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.119 (sd 0.0056765) 

Confusion Matrix: 

         observed

predicted    1    0

        1  306  568

        0   47 4083

AUC                          :  0.94675  (sd 0.0090023) 

Percent Correctly Classified :  87.71026  (sd 1.6104) 

Sensitivity                  :  0.8671429  (sd 0.049256) 

Specificity                  :  0.8778795  (sd 0.017753) 

Kappa                        :  0.444858  (sd 0.042765) 

True Skill Statistic         :  0.7450223  (sd 0.048504) 

   Calibration Statistics 

Intercept (general calibration)                            :  0.01473715  (sd 0.28396) 

Slope   (direction and variation in fit)                   :  1.000783  (sd 0.13166) 

Testa0b1 (overall reliability of predictors)               :  0.6486042  (sd 0.2832) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.6212029  (sd 0.27908) 

Testb1|a (refinement given correct calibration)            :  0.5285948  (sd 0.23744) 

Total time = 0.57 min

Random Forest Modeling Results

Data:

C:\Withrow\Workspace\Model\rf_1\CovariateCorrelationOutputMDS_initial.csv

n(pres)                 =      353

n(abs)                  =      4651

n covariates considered =      7

total time for model fitting = 0.12min

Settings:

random seed used                       : 19680828

n covariates considered at each split  : 2

n trees                                : 1000

Relative performance of predictors in final model:

                   0       1 MeanDecreaseAccuracy MeanDecreaseGini

bio02        51.2423 61.7172              71.5781          64.6245

bio04        34.2271 97.8122              62.6355          98.5002

bio17        26.3046 37.5794              47.9493          45.4199

bio08         8.2788 78.4877              42.6005          66.1096

di_all_model 27.7573 31.4360              41.9453          39.1254

bio06        31.6245 29.0094              41.0153          47.7090

bio09        12.6181 39.9307              37.2954          52.7073

============================================================ 

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.6770872 

NULL Deviance                : 0.51009

Fit Deviance                 : 0.2389

Explained Deviance           : 0.27118

Percent Deviance Explained   : 53.164

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.1 

Confusion Matrix: 

         observed

predicted    1    0

        1  315  517

        0   38 4134

AUC                          :  0.9604 

Percent Correctly Classified :  88.90887 

Sensitivity                  :  0.8923513 

Specificity                  :  0.8888411 

Kappa                        :  0.4801499 

True Skill Statistic         :  0.7811924 

   Calibration Statistics 

Intercept (general calibration)                            :  0.0167778 

Slope   (direction and variation in fit)                   :  1.069352 

Testa0b1 (overall reliability of predictors)               :  0.2489207 

Testa0|b1(incorrect calibration given correct refinement)  :  0.4765319 

Testb1|a (refinement given correct calibration)            :  0.1315216 

============================================================ 

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6685274  (sd 0.044275) 

NULL Deviance                : 0.51007 (sd 0.0046687)

Fit Deviance                 : 0.24475 (sd 0.026463)

Explained Deviance           : 0.26532 (sd 0.026088)

Percent Deviance Explained   : 52.02 (sd 5.123)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.481 (sd 0.0061464) 

Confusion Matrix: 

         observed

predicted    1    0

        1  194   88

        0  159 4563

AUC                          :  0.95656  (sd 0.012676) 

Percent Correctly Classified :  95.06423  (sd 0.85153) 

Sensitivity                  :  0.5498413  (sd 0.074661) 

Specificity                  :  0.9810803  (sd 0.0066283) 

Kappa                        :  0.5845288  (sd 0.072385) 

True Skill Statistic         :  0.5309216  (sd 0.075406) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.03587436  (sd 0.26353) 

Slope   (direction and variation in fit)                   :  1.003453  (sd 0.17205) 

Testa0b1 (overall reliability of predictors)               :  0.481483  (sd 0.23926) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.5504052  (sd 0.20097) 

Testb1|a (refinement given correct calibration)            :  0.3678857  (sd 0.26212) 

Total time = 2.07 min