Species 065: Juniperus osteosperma

General Linear Models

Boosted Regression Trees

Generalized Linear Model Results

Data:

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

n(pres)                 =      520

n(abs)                  =      4495

n covariates considered =      8

total time for model fitting = 2.34min

Settings:

model family          : binomial

simplification method : AIC

Results:

number covariates in final model   : 35

Call:

glm(formula = response ~ I(bio15^2) + I(bio11^2) + di_all_model + 

    bio18 + I(bio18^2) + bio15 + bio08 + I(di_all_model^2) + 

    bio02 + bio11 + I(bio19^2) + bio19 + I(bio08^2) + bio04 + 

    I(bio04^2) + I(bio02^2) + bio18:bio08 + di_all_model:bio15 + 

    bio18:bio15 + bio18:bio02 + bio08:bio02 + bio18:bio11 + bio08:bio11 + 

    di_all_model:bio08 + bio08:bio19 + bio18:bio19 + bio11:bio04 + 

    bio15:bio11 + bio18:bio04 + bio19:bio04 + bio02:bio19 + bio02:bio04 + 

    bio02:bio11 + bio15:bio02 + di_all_model:bio04, family = model.family, 

    data = dat, weights = weight, na.action = "na.exclude")

Deviance Residuals: 

    Min       1Q   Median       3Q      Max  

-2.1565  -0.2697  -0.0432  -0.0004   3.6082  

Coefficients:

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

(Intercept)         3.261e+01  2.683e+01   1.215 0.224199    

I(bio15^2)          2.699e-03  4.379e-04   6.163 7.14e-10 ***

I(bio11^2)         -9.264e-04  1.036e-04  -8.946  < 2e-16 ***

di_all_model       -4.006e-02  6.320e-02  -0.634 0.526194    

bio18               2.409e-01  6.345e-02   3.796 0.000147 ***

I(bio18^2)         -3.996e-04  1.194e-04  -3.347 0.000818 ***

bio15              -2.996e-01  8.438e-02  -3.550 0.000385 ***

bio08               5.429e-02  2.133e-02   2.545 0.010932 *  

I(di_all_model^2)   7.312e-04  1.647e-04   4.441 8.96e-06 ***

bio02              -7.467e-01  1.755e-01  -4.254 2.10e-05 ***

bio11               2.226e-01  7.492e-02   2.971 0.002968 ** 

I(bio19^2)         -1.918e-04  6.541e-05  -2.933 0.003362 ** 

bio19               1.973e-02  6.422e-02   0.307 0.758669    

I(bio08^2)          1.007e-04  2.850e-05   3.533 0.000411 ***

bio04               5.962e-03  3.913e-03   1.524 0.127579    

I(bio04^2)         -7.768e-07  1.878e-07  -4.136 3.54e-05 ***

I(bio02^2)          5.740e-04  3.371e-04   1.703 0.088649 .  

bio18:bio08        -3.744e-04  9.493e-05  -3.944 8.03e-05 ***

di_all_model:bio15  8.041e-04  4.333e-04   1.856 0.063515 .  

bio18:bio15        -2.007e-03  3.280e-04  -6.117 9.51e-10 ***

bio18:bio02         1.129e-03  2.615e-04   4.318 1.58e-05 ***

bio08:bio02        -4.132e-04  1.195e-04  -3.457 0.000546 ***

bio18:bio11         9.992e-04  1.696e-04   5.892 3.82e-09 ***

bio08:bio11        -2.674e-04  5.887e-05  -4.542 5.57e-06 ***

di_all_model:bio08  2.671e-04  9.417e-05   2.836 0.004562 ** 

bio08:bio19         2.014e-04  7.056e-05   2.855 0.004308 ** 

bio18:bio19        -5.347e-04  1.108e-04  -4.826 1.39e-06 ***

bio11:bio04        -4.094e-05  7.057e-06  -5.802 6.57e-09 ***

bio15:bio11        -9.924e-04  2.776e-04  -3.575 0.000350 ***

bio18:bio04        -1.789e-05  5.855e-06  -3.055 0.002247 ** 

bio19:bio04        -1.022e-05  5.556e-06  -1.840 0.065740 .  

bio02:bio19         8.102e-04  2.159e-04   3.752 0.000175 ***

bio02:bio04         4.972e-05  1.340e-05   3.710 0.000208 ***

bio02:bio11         4.286e-04  2.597e-04   1.650 0.098851 .  

bio15:bio02         9.668e-04  4.955e-04   1.951 0.051063 .  

di_all_model:bio04 -1.278e-05  6.790e-06  -1.881 0.059925 .  

---

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

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 3341.1  on 5014  degrees of freedom

Residual deviance: 1606.7  on 4979  degrees of freedom

AIC: 1678.7

Number of Fisher Scoring iterations: 10

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.7030271 

NULL Deviance                : 0.66623

Fit Deviance                 : 0.32038

Explained Deviance           : 0.34585

Percent Deviance Explained   : 51.912

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.11 

Confusion Matrix: 

         observed

predicted    1    0

        1  452  630

        0   68 3865

AUC                          :  0.9471 

Percent Correctly Classified :  86.08175 

Sensitivity                  :  0.8692308 

Specificity                  :  0.8598443 

Kappa                        :  0.493328 

True Skill Statistic         :  0.729075 

   Calibration Statistics 

Intercept (general calibration)                            :  0.0002098733 

Slope   (direction and variation in fit)                   :  1.000442 

Testa0b1 (overall reliability of predictors)               :  0.9999394 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9974397 

Testb1|a (refinement given correct calibration)            :  0.9916011 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6856128  (sd 0.043598) 

NULL Deviance                : 0.66623 (sd 0.0006313)

Fit Deviance                 : 0.35064 (sd 0.050567)

Explained Deviance           : 0.31559 (sd 0.050411)

Percent Deviance Explained   : 47.371 (sd 7.5805)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.121 (sd 0.0056765) 

Confusion Matrix: 

         observed

predicted    1    0

        1  446  616

        0   74 3879

AUC                          :  0.93716  (sd 0.021028) 

Percent Correctly Classified :  86.24165  (sd 1.599) 

Sensitivity                  :  0.8576923  (sd 0.072069) 

Specificity                  :  0.8629674  (sd 0.015654) 

Kappa                        :  0.4937025  (sd 0.049018) 

True Skill Statistic         :  0.7206597  (sd 0.073978) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.05538643  (sd 0.25807) 

Slope   (direction and variation in fit)                   :  0.9693207  (sd 0.22132) 

Testa0b1 (overall reliability of predictors)               :  0.3454815  (sd 0.3235) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.4997207  (sd 0.29817) 

Testb1|a (refinement given correct calibration)            :  0.3124696  (sd 0.32578) 

Total time = 17 min

Boosted Regression Tree Modeling Results

Data:

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

n(pres)                 =      520

n(abs)                  =      4495

n covariates considered =      8

total time for model fitting = 0.45min

Settings:

random seed used             : 19680828

tree complexity              : 9

learning rate                : 0.0645

n(trees)                     : 300

model simplification         : cross-validation

n folds                      : 3

n covariates in final model  : 3

Relative influence of predictors in final model:

   Var  rel.inf

 bio11 40.11528

 bio18 32.84515

 bio15 27.03957

Important interactions in final model:

 v1 name1 v2 name2

  3 bio18  2 bio15

  2 bio15  1 bio11

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.8233794 

NULL Deviance                : 0.66623

Fit Deviance                 : 0.2196

Explained Deviance           : 0.44663

Percent Deviance Explained   : 67.038

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.17 

Confusion Matrix: 

         observed

predicted    1    0

        1  481  312

        0   39 4183

AUC                          :  0.9812 

Percent Correctly Classified :  93.001 

Sensitivity                  :  0.925 

Specificity                  :  0.9305895 

Kappa                        :  0.6943971 

True Skill Statistic         :  0.8555895 

   Calibration Statistics 

Intercept (general calibration)                            :  0.4750768 

Slope   (direction and variation in fit)                   :  1.533945 

Testa0b1 (overall reliability of predictors)               :  0 

Testa0|b1(incorrect calibration given correct refinement)  :  0.84262 

Testb1|a (refinement given correct calibration)            :  0 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6647328  (sd 0.059217) 

NULL Deviance                : 0.66623 (sd 0.0006313)

Fit Deviance                 : 0.3576 (sd 0.055933)

Explained Deviance           : 0.30863 (sd 0.055855)

Percent Deviance Explained   : 46.326 (sd 8.3913)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.191 (sd 0.024244) 

Confusion Matrix: 

         observed

predicted    1    0

        1  394  324

        0  126 4171

AUC                          :  0.93472  (sd 0.022813) 

Percent Correctly Classified :  91.02714  (sd 1.9787) 

Sensitivity                  :  0.7576923  (sd 0.048987) 

Specificity                  :  0.9279237  (sd 0.019855) 

Kappa                        :  0.5897263  (sd 0.069781) 

True Skill Statistic         :  0.685616  (sd 0.057645) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.1208739  (sd 0.24402) 

Slope   (direction and variation in fit)                   :  0.8564017  (sd 0.091355) 

Testa0b1 (overall reliability of predictors)               :  0.2916589  (sd 0.26265) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.4206643  (sd 0.30744) 

Testb1|a (refinement given correct calibration)            :  0.297458  (sd 0.32135) 

Total time = 5.28 min

MARS

Random Forests

MARS Model Results

Data:

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

n(pres)                 =      520

n(abs)                  =      4495

n covariates considered =      8

total time for model fitting = 0.05min

Settings:

random seed used             : 123

mars degree                  : 1

mars penalty                 : 2

Summary of Model:

             nsubsets   gcv    rss

bio18              14 100.0  100.0

di_all_model       11  69.6   70.1

bio19              10  62.6   63.2

bio08               9  56.8   57.5

bio15               7  43.4   44.3

bio11               6  35.2   36.3

bio04               4  19.0   20.6

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.6463863 

NULL Deviance                : 0.66623

Fit Deviance                 : 0.36857

Explained Deviance           : 0.29766

Percent Deviance Explained   : 44.678

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.12 

Confusion Matrix: 

         observed

predicted    1    0

        1  447  645

        0   73 3850

AUC                          :  0.926 

Percent Correctly Classified :  85.68295 

Sensitivity                  :  0.8596154 

Specificity                  :  0.8565072 

Kappa                        :  0.4817918 

True Skill Statistic         :  0.7161226 

   Calibration Statistics 

Intercept (general calibration)                            :  8.728224e-05 

Slope   (direction and variation in fit)                   :  1.000239 

Testa0b1 (overall reliability of predictors)               :  0.9999755 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9975854 

Testb1|a (refinement given correct calibration)            :  0.994968 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6347533  (sd 0.035314) 

NULL Deviance                : 0.66623 (sd 0.0006313)

Fit Deviance                 : 0.37855 (sd 0.026389)

Explained Deviance           : 0.28768 (sd 0.026228)

Percent Deviance Explained   : 43.182 (sd 3.9483)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.12 (sd 0) 

Confusion Matrix: 

         observed

predicted    1    0

        1  443  658

        0   77 3837

AUC                          :  0.92256  (sd 0.012522) 

Percent Correctly Classified :  85.3444  (sd 1.1186) 

Sensitivity                  :  0.8519231  (sd 0.049695) 

Specificity                  :  0.8536202  (sd 0.01435) 

Kappa                        :  0.4725389  (sd 0.027068) 

True Skill Statistic         :  0.7055433  (sd 0.044228) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.01547108  (sd 0.18734) 

Slope   (direction and variation in fit)                   :  0.99035  (sd 0.10287) 

Testa0b1 (overall reliability of predictors)               :  0.5208512  (sd 0.22574) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.4382947  (sd 0.21193) 

Testb1|a (refinement given correct calibration)            :  0.5455324  (sd 0.31599) 

Total time = 0.97 min

Random Forest Modeling Results

Data:

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

n(pres)                 =      520

n(abs)                  =      4495

n covariates considered =      8

total time for model fitting = 0.15min

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

bio18        64.8150 63.2378              91.9578         108.7301

bio11        60.9294 58.2124              77.7201          82.5189

di_all_model 53.0141 53.3990              69.2890          56.9028

bio08        58.1255 46.0096              68.4943          80.2508

bio19        46.4798 60.4473              65.6052          74.3309

bio15        41.9454 83.3291              59.6827          84.4569

bio04        43.3873 39.9481              56.9929          58.8765

bio02        39.0799 24.7022              45.5813          43.1684

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.7591163 

NULL Deviance                : 0.66623

Fit Deviance                 : 0.2741

Explained Deviance           : 0.39212

Percent Deviance Explained   : 58.857

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.14 

Confusion Matrix: 

         observed

predicted    1    0

        1  467  444

        0   53 4051

AUC                          :  0.9633 

Percent Correctly Classified :  90.08973 

Sensitivity                  :  0.8980769 

Specificity                  :  0.9012236 

Kappa                        :  0.5998643 

True Skill Statistic         :  0.7993005 

   Calibration Statistics 

Intercept (general calibration)                            :  0.2626816 

Slope   (direction and variation in fit)                   :  1.247746 

Testa0b1 (overall reliability of predictors)               :  2.419414e-07 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9157712 

Testb1|a (refinement given correct calibration)            :  3.411814e-08 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.7551031  (sd 0.044561) 

NULL Deviance                : 0.66623 (sd 0.0006313)

Fit Deviance                 : 0.27963 (sd 0.037447)

Explained Deviance           : 0.38659 (sd 0.037544)

Percent Deviance Explained   : 58.027 (sd 5.6297)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.4815 (sd 0.0078351) 

Confusion Matrix: 

         observed

predicted    1    0

        1  328   74

        0  192 4421

AUC                          :  0.96218  (sd 0.014262) 

Percent Correctly Classified :  94.6965  (sd 1.1973) 

Sensitivity                  :  0.6307692  (sd 0.087799) 

Specificity                  :  0.9835486  (sd 0.01104) 

Kappa                        :  0.6816986  (sd 0.072496) 

True Skill Statistic         :  0.6143179  (sd 0.086221) 

   Calibration Statistics 

Intercept (general calibration)                            :  0.2658576  (sd 0.40992) 

Slope   (direction and variation in fit)                   :  1.255609  (sd 0.27496) 

Testa0b1 (overall reliability of predictors)               :  0.2979787  (sd 0.31735) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.4868493  (sd 0.26599) 

Testb1|a (refinement given correct calibration)            :  0.2426459  (sd 0.27964) 

Total time = 3.89 min