Species 117: Pinus aristata

General Linear Models

Boosted Regression Trees

Generalized Linear Model Results

Data:

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

n(pres)                 =      338

n(abs)                  =      4667

n covariates considered =      7

total time for model fitting = 0.57min

Settings:

model family          : binomial

simplification method : AIC

Results:

number covariates in final model   : 18

Call:

glm(formula = response ~ bio15 + I(bio08^2) + bio03 + I(bio03^2) + 

    I(bio14^2) + bio16 + I(bio16^2) + bio09 + I(bio09^2) + I(bio15^2) + 

    I(di_all_model^2) + di_all_model + bio14 + bio03:bio09 + 

    bio15:bio09 + bio15:bio03 + bio03:bio16 + bio03:bio14, family = model.family, 

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

Deviance Residuals: 

    Min       1Q   Median       3Q      Max  

-2.3393  -0.0270  -0.0015   0.0000   3.6417  

Coefficients:

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

(Intercept)       -7.603e+01  5.555e+01  -1.369 0.171077    

bio15             -1.776e+00  4.272e-01  -4.157 3.22e-05 ***

I(bio08^2)        -1.131e-03  1.529e-04  -7.396 1.41e-13 ***

bio03              3.099e+00  1.761e+00   1.760 0.078433 .  

I(bio03^2)        -3.642e-02  1.449e-02  -2.513 0.011970 *  

I(bio14^2)        -1.501e-02  6.743e-03  -2.227 0.025980 *  

bio16              8.227e-02  1.469e-02   5.599 2.16e-08 ***

I(bio16^2)        -1.989e-05  3.969e-06  -5.011 5.41e-07 ***

bio09              6.502e-01  1.483e-01   4.384 1.16e-05 ***

I(bio09^2)        -9.832e-04  1.828e-04  -5.379 7.50e-08 ***

I(bio15^2)        -5.110e-03  1.369e-03  -3.732 0.000190 ***

I(di_all_model^2) -6.364e-04  2.501e-04  -2.545 0.010928 *  

di_all_model       3.407e-02  1.492e-02   2.284 0.022393 *  

bio14             -2.290e+00  1.265e+00  -1.810 0.070250 .  

bio03:bio09       -1.330e-02  2.950e-03  -4.511 6.46e-06 ***

bio15:bio09        4.217e-03  8.342e-04   5.055 4.29e-07 ***

bio15:bio03        3.926e-02  1.050e-02   3.739 0.000185 ***

bio03:bio16       -1.234e-03  3.349e-04  -3.685 0.000229 ***

bio03:bio14        5.279e-02  2.611e-02   2.021 0.043240 *  

---

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

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 2474.56  on 5004  degrees of freedom

Residual deviance:  752.65  on 4986  degrees of freedom

AIC: 790.65

Number of Fisher Scoring iterations: 12

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.7983906 

NULL Deviance                : 0.49442

Fit Deviance                 : 0.15038

Explained Deviance           : 0.34404

Percent Deviance Explained   : 69.585

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.1 

Confusion Matrix: 

         observed

predicted    1    0

        1  317  291

        0   21 4376

AUC                          :  0.9855 

Percent Correctly Classified :  93.76623 

Sensitivity                  :  0.9378698 

Specificity                  :  0.9376473 

Kappa                        :  0.6388389 

True Skill Statistic         :  0.8755171 

   Calibration Statistics 

Intercept (general calibration)                            :  6.261828e-05 

Slope   (direction and variation in fit)                   :  1.001291 

Testa0b1 (overall reliability of predictors)               :  0.9997302 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9963086 

Testb1|a (refinement given correct calibration)            :  0.9818369 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.7871675  (sd 0.03654) 

NULL Deviance                : 0.4944 (sd 0.0037479)

Fit Deviance                 : 0.1665 (sd 0.021245)

Explained Deviance           : 0.3279 (sd 0.021333)

Percent Deviance Explained   : 66.323 (sd 4.2938)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.113 (sd 0.0082327) 

Confusion Matrix: 

         observed

predicted    1    0

        1  313  290

        0   25 4377

AUC                          :  0.98224  (sd 0.0046378) 

Percent Correctly Classified :  93.70575  (sd 0.7479) 

Sensitivity                  :  0.9262923  (sd 0.062348) 

Specificity                  :  0.9378558  (sd 0.010387) 

Kappa                        :  0.6340848  (sd 0.029787) 

True Skill Statistic         :  0.8641481  (sd 0.055758) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.07242661  (sd 0.251) 

Slope   (direction and variation in fit)                   :  0.9506497  (sd 0.20801) 

Testa0b1 (overall reliability of predictors)               :  0.4660885  (sd 0.33659) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.6090238  (sd 0.37101) 

Testb1|a (refinement given correct calibration)            :  0.3604657  (sd 0.26815) 

Total time = 5.62 min

Boosted Regression Tree Modeling Results

Data:

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

n(pres)                 =      338

n(abs)                  =      4667

n covariates considered =      7

total time for model fitting = 0.28min

Settings:

random seed used             : 19680828

tree complexity              : 9

learning rate                : 0.0524

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

 bio08 39.42277

 bio16 37.97568

 bio14 22.60156

Important interactions in final model:

 v1 name1 v2 name2

  3 bio16  2 bio14

  3 bio16  1 bio08

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.9053775 

NULL Deviance                : 0.49442

Fit Deviance                 : 0.085625

Explained Deviance           : 0.40879

Percent Deviance Explained   : 82.682

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.21 

Confusion Matrix: 

         observed

predicted    1    0

        1  330  117

        0    8 4550

AUC                          :  0.9969 

Percent Correctly Classified :  97.5025 

Sensitivity                  :  0.9763314 

Specificity                  :  0.9749304 

Kappa                        :  0.8274972 

True Skill Statistic         :  0.9512617 

   Calibration Statistics 

Intercept (general calibration)                            :  0.3534495 

Slope   (direction and variation in fit)                   :  1.826928 

Testa0b1 (overall reliability of predictors)               :  1.110223e-16 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9829219 

Testb1|a (refinement given correct calibration)            :  0 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6919071  (sd 0.085379) 

NULL Deviance                : 0.4944 (sd 0.0037479)

Fit Deviance                 : 0.21897 (sd 0.057948)

Explained Deviance           : 0.27543 (sd 0.057614)

Percent Deviance Explained   : 55.718 (sd 11.647)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.218 (sd 0.028597) 

Confusion Matrix: 

         observed

predicted    1    0

        1  261  199

        0   77 4468

AUC                          :  0.97054  (sd 0.015841) 

Percent Correctly Classified :  94.48468  (sd 1.0286) 

Sensitivity                  :  0.7723708  (sd 0.081728) 

Specificity                  :  0.9573511  (sd 0.0075137) 

Kappa                        :  0.6249441  (sd 0.06828) 

True Skill Statistic         :  0.7297219  (sd 0.084581) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.2036997  (sd 0.21446) 

Slope   (direction and variation in fit)                   :  0.7741092  (sd 0.12721) 

Testa0b1 (overall reliability of predictors)               :  0.2248248  (sd 0.27463) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.430406  (sd 0.27545) 

Testb1|a (refinement given correct calibration)            :  0.2160034  (sd 0.29747) 

Total time = 3.41 min

MARS

Random Forests

MARS Model Results

Data:

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

n(pres)                 =      338

n(abs)                  =      4667

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

bio16       14 100.0  100.0

bio15       12  63.8   64.4

bio03        9  46.1   46.8

bio08        9  41.1   41.9

bio14        5  30.3   31.0

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.7712845 

NULL Deviance                : 0.49442

Fit Deviance                 : 0.17076

Explained Deviance           : 0.32366

Percent Deviance Explained   : 65.463

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.09 

Confusion Matrix: 

         observed

predicted    1    0

        1  310  346

        0   28 4321

AUC                          :  0.9803 

Percent Correctly Classified :  92.52747 

Sensitivity                  :  0.9171598 

Specificity                  :  0.9258624 

Kappa                        :  0.5869217 

True Skill Statistic         :  0.8430222 

   Calibration Statistics 

Intercept (general calibration)                            :  0.0002155923 

Slope   (direction and variation in fit)                   :  1.000864 

Testa0b1 (overall reliability of predictors)               :  0.9998462 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9964841 

Testb1|a (refinement given correct calibration)            :  0.986453 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.763516  (sd 0.047392) 

NULL Deviance                : 0.4944 (sd 0.0037479)

Fit Deviance                 : 0.19223 (sd 0.052836)

Explained Deviance           : 0.30218 (sd 0.052314)

Percent Deviance Explained   : 61.13 (sd 10.604)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.086 (sd 0.0069921) 

Confusion Matrix: 

         observed

predicted    1    0

        1  306  365

        0   32 4302

AUC                          :  0.97614  (sd 0.0086034) 

Percent Correctly Classified :  92.0675  (sd 0.95896) 

Sensitivity                  :  0.9053476  (sd 0.05506) 

Specificity                  :  0.921787  (sd 0.012069) 

Kappa                        :  0.5686108  (sd 0.032607) 

True Skill Statistic         :  0.8271346  (sd 0.049091) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.04199831  (sd 0.22332) 

Slope   (direction and variation in fit)                   :  0.9739026  (sd 0.21035) 

Testa0b1 (overall reliability of predictors)               :  0.5191839  (sd 0.31717) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.5037387  (sd 0.31353) 

Testb1|a (refinement given correct calibration)            :  0.4482182  (sd 0.29433) 

Total time = 0.57 min

Random Forest Modeling Results

Data:

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

n(pres)                 =      338

n(abs)                  =      4667

n covariates considered =      7

total time for model fitting = 0.11min

Settings:

random seed used                       : 19680828

n covariates considered at each split  : 6

n trees                                : 1000

Relative performance of predictors in final model:

                   0        1 MeanDecreaseAccuracy MeanDecreaseGini

bio08        51.3092 124.9080             121.7339         112.6242

bio16        45.0326 177.9007              92.5364          97.2990

bio14        57.8836  60.7307              74.7498          59.9791

bio03         4.4086  95.2890              74.2806          35.9832

bio09        34.5013  36.5680              47.8160          34.8123

di_all_model 14.5784  17.7554              24.0374          22.3650

bio15        16.6852  18.2082              20.9747          34.5487

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.7867627 

NULL Deviance                : 0.49442

Fit Deviance                 : 0.16227

Explained Deviance           : 0.33214

Percent Deviance Explained   : 67.179

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.11 

Confusion Matrix: 

         observed

predicted    1    0

        1  315  282

        0   23 4385

AUC                          :  0.9826 

Percent Correctly Classified :  93.90609 

Sensitivity                  :  0.9319527 

Specificity                  :  0.9395757 

Kappa                        :  0.6430102 

True Skill Statistic         :  0.8715284 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.09742587 

Slope   (direction and variation in fit)                   :  0.9964476 

Testa0b1 (overall reliability of predictors)               :  0.5977118 

Testa0|b1(incorrect calibration given correct refinement)  :  0.3117882 

Testb1|a (refinement given correct calibration)            :  0.9372018 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.7964653  (sd 0.043655) 

NULL Deviance                : 0.4944 (sd 0.0037479)

Fit Deviance                 : 0.15275 (sd 0.025124)

Explained Deviance           : 0.34165 (sd 0.023392)

Percent Deviance Explained   : 69.12 (sd 4.9546)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.489 (sd 0.0065828) 

Confusion Matrix: 

         observed

predicted    1    0

        1  233   60

        0  105 4607

AUC                          :  0.98536  (sd 0.0050213) 

Percent Correctly Classified :  96.70266  (sd 0.74067) 

Sensitivity                  :  0.6893048  (sd 0.078855) 

Specificity                  :  0.98714  (sd 0.0044195) 

Kappa                        :  0.7197897  (sd 0.06598) 

True Skill Statistic         :  0.6764448  (sd 0.07987) 

   Calibration Statistics 

Intercept (general calibration)                            :  0.01655369  (sd 0.19037) 

Slope   (direction and variation in fit)                   :  1.093414  (sd 0.14414) 

Testa0b1 (overall reliability of predictors)               :  0.5884259  (sd 0.27361) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.6470398  (sd 0.32698) 

Testb1|a (refinement given correct calibration)            :  0.4830021  (sd 0.24996) 

Total time = 1.76 min