Species 104: Pinus balfouriana

General Linear Models

Boosted Regression Trees

Generalized Linear Model Results

Data:

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

n(pres)                 =      22

n(abs)                  =      4980

n covariates considered =      7

total time for model fitting = 0.84min

Settings:

model family          : binomial

simplification method : AIC

Results:

number covariates in final model   : 9

Call:

glm(formula = response ~ I(bio05^2) + I(bio08^2) + bio03 + di_all_model + 

    bio05 + I(bio03^2) + bio18 + I(bio18^2) + bio05:bio18, family = model.family, 

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

Deviance Residuals: 

   Min      1Q  Median      3Q     Max  

-1.416   0.000   0.000   0.000   3.084  

Coefficients:

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

(Intercept)  -26.283225 229.881854  -0.114  0.90897   

I(bio05^2)    -0.014129   0.004413  -3.202  0.00137 **

I(bio08^2)    -0.003214   0.001744  -1.843  0.06529 . 

bio03        -50.288732  15.486088  -3.247  0.00116 **

di_all_model  -1.297598  24.399862  -0.053  0.95759   

bio05          7.894573   2.472434   3.193  0.00141 **

I(bio03^2)     0.569354   0.178388   3.192  0.00141 **

bio18         11.419132   3.934624   2.902  0.00371 **

I(bio18^2)    -0.030258   0.011756  -2.574  0.01006 * 

bio05:bio18   -0.037741   0.012785  -2.952  0.00316 **

---

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

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 282.671  on 5001  degrees of freedom

Residual deviance:  59.156  on 4992  degrees of freedom

AIC: 79.156

Number of Fisher Scoring iterations: 25

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.7798042 

NULL Deviance                : 0.056512

Fit Deviance                 : 0.011827

Explained Deviance           : 0.044685

Percent Deviance Explained   : 79.072

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.01 

Confusion Matrix: 

         observed

predicted    1    0

        1   21   60

        0    1 4920

AUC                          :  0.9985 

Percent Correctly Classified :  98.78049 

Sensitivity                  :  0.9545455 

Specificity                  :  0.9879518 

Kappa                        :  0.4036416 

True Skill Statistic         :  0.9424973 

   Calibration Statistics 

Intercept (general calibration)                            :  0.01290939 

Slope   (direction and variation in fit)                   :  1.020635 

Testa0b1 (overall reliability of predictors)               :  0.9945826 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9867966 

Testb1|a (refinement given correct calibration)            :  0.9180347 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6134964  (sd 0.24532) 

NULL Deviance                : 0.056371 (sd 0.0088872)

Fit Deviance                 : 0.025579 (sd 0.01767)

Explained Deviance           : 0.030792 (sd 0.018458)

Percent Deviance Explained   : 54.374 (sd 33.099)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.016 (sd 0.0084327) 

Confusion Matrix: 

         observed

predicted    1    0

        1   19   53

        0    3 4927

AUC                          :  0.99569  (sd 0.0042751) 

Percent Correctly Classified :  98.8804  (sd 0.37919) 

Sensitivity                  :  0.8666667  (sd 0.21943) 

Specificity                  :  0.9893574  (sd 0.0032859) 

Kappa                        :  0.4110198  (sd 0.15688) 

True Skill Statistic         :  0.8560241  (sd 0.22074) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.1927401  (sd 7.0216) 

Slope   (direction and variation in fit)                   :  4.276981  (sd 6.8889) 

Testa0b1 (overall reliability of predictors)               :  0.265059  (sd 0.24216) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.4517072  (sd 0.30271) 

Testb1|a (refinement given correct calibration)            :  0.2826364  (sd 0.30997) 

Total time = 9.63 min

Boosted Regression Tree Modeling Results

Data:

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

n(pres)                 =      22

n(abs)                  =      4980

n covariates considered =      7

total time for model fitting = 0.52min

Settings:

random seed used             : 19680828

tree complexity              : 9

learning rate                : 0.0316

n(trees)                     : 300

model simplification         : cross-validation

n folds                      : 3

n covariates in final model  : 2

Relative influence of predictors in final model:

   Var  rel.inf

 bio05 72.56575

 bio07 27.43425

Important interactions in final model:

NULL

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.877026 

NULL Deviance                : 0.056512

Fit Deviance                 : 0.0079283

Explained Deviance           : 0.048583

Percent Deviance Explained   : 85.97

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.31 

Confusion Matrix: 

         observed

predicted    1    0

        1   22    7

        0    0 4973

AUC                          :  0.9998 

Percent Correctly Classified :  99.86006 

Sensitivity                  :  1 

Specificity                  :  0.9985944 

Kappa                        :  0.8620551 

True Skill Statistic         :  0.9985944 

   Calibration Statistics 

Intercept (general calibration)                            :  1.900048 

Slope   (direction and variation in fit)                   :  4.378436 

Testa0b1 (overall reliability of predictors)               :  0.0001327132 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9860775 

Testb1|a (refinement given correct calibration)            :  2.384752e-05 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.4306442  (sd 0.18019) 

NULL Deviance                : 0.056371 (sd 0.0088872)

Fit Deviance                 : 0.033264 (sd 0.022797)

Explained Deviance           : 0.023107 (sd 0.019275)

Percent Deviance Explained   : 42.57 (sd 32.949)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.3085 (sd 0.053492) 

Confusion Matrix: 

         observed

predicted    1    0

        1    7   16

        0   15 4964

AUC                          :  0.99265  (sd 0.0043046) 

Percent Correctly Classified :  99.38032  (sd 0.30451) 

Sensitivity                  :  0.3166667  (sd 0.3282) 

Specificity                  :  0.9967871  (sd 0.00254) 

Kappa                        :  0.2891605  (sd 0.27718) 

True Skill Statistic         :  0.3134538  (sd 0.32837) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.1516136  (sd 0.94283) 

Slope   (direction and variation in fit)                   :  0.866534  (sd 0.66429) 

Testa0b1 (overall reliability of predictors)               :  0.4139148  (sd 0.28264) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.5880176  (sd 0.29544) 

Testb1|a (refinement given correct calibration)            :  0.3447775  (sd 0.25208) 

Total time = 6.11 min

MARS

Random Forests

MARS Model Results

Data:

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

n(pres)                 =      22

n(abs)                  =      4980

n covariates considered =      7

total time for model fitting = 0.08min

Settings:

random seed used             : 123

mars degree                  : 1

mars penalty                 : 2

Summary of Model:

             nsubsets   gcv    rss

bio05               3 100.0  100.0

bio18               2  11.6   14.5

di_all_model        1   2.3    6.6

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.3948995 

NULL Deviance                : 0.056512

Fit Deviance                 : 0.026492

Explained Deviance           : 0.030019

Percent Deviance Explained   : 53.121

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.01 

Confusion Matrix: 

         observed

predicted    1    0

        1   22   95

        0    0 4885

AUC                          :  0.9934 

Percent Correctly Classified :  98.10076 

Sensitivity                  :  1 

Specificity                  :  0.9809237 

Kappa                        :  0.3114486 

True Skill Statistic         :  0.9809237 

   Calibration Statistics 

Intercept (general calibration)                            :  0.008863278 

Slope   (direction and variation in fit)                   :  1.004971 

Testa0b1 (overall reliability of predictors)               :  0.9989913 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9900237 

Testb1|a (refinement given correct calibration)            :  0.9655799 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.2989966  (sd 0.26293) 

NULL Deviance                : 0.056371 (sd 0.0088872)

Fit Deviance                 : 0.03501 (sd 0.014875)

Explained Deviance           : 0.021362 (sd 0.012474)

Percent Deviance Explained   : 38.443 (sd 23.041)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.015 (sd 0.0097183) 

Confusion Matrix: 

         observed

predicted    1    0

        1   19  101

        0    3 4879

AUC                          :  0.95106  (sd 0.086234) 

Percent Correctly Classified :  97.92064  (sd 0.87062) 

Sensitivity                  :  0.8666667  (sd 0.21943) 

Specificity                  :  0.9797189  (sd 0.0088521) 

Kappa                        :  0.2748617  (sd 0.099064) 

True Skill Statistic         :  0.8463855  (sd 0.21828) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.4525424  (sd 0.95847) 

Slope   (direction and variation in fit)                   :  0.9377088  (sd 0.37863) 

Testa0b1 (overall reliability of predictors)               :  0.5723161  (sd 0.27809) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.6741344  (sd 0.24524) 

Testb1|a (refinement given correct calibration)            :  0.431336  (sd 0.2739) 

Total time = 0.71 min

Random Forest Modeling Results

Data:

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

n(pres)                 =      22

n(abs)                  =      4980

n covariates considered =      7

total time for model fitting = 0.21min

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

bio03         23.8390  8.6178              25.2519           3.5183

bio05         12.8591 18.5092              17.2204           6.4428

bio09         12.3201  1.6650              14.5274           4.1662

bio07          9.7598 17.2222              13.9249           4.6797

bio08          6.8631  9.2686              11.6992           4.6336

bio18         -1.4684 12.0261               5.3074           3.2858

di_all_model -16.5676 20.0769              -6.0736           0.5891

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.5874086 

NULL Deviance                : 0.056512

Fit Deviance                 : 0.025957

Explained Deviance           : 0.030554

Percent Deviance Explained   : 54.067

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.01 

Confusion Matrix: 

         observed

predicted    1    0

        1   20  121

        0    2 4859

AUC                          :  0.9719 

Percent Correctly Classified :  97.54098 

Sensitivity                  :  0.9090909 

Specificity                  :  0.9757028 

Kappa                        :  0.2396128 

True Skill Statistic         :  0.8847937 

   Calibration Statistics 

Intercept (general calibration)                            :  0.2219665 

Slope   (direction and variation in fit)                   :  1.32731 

Testa0b1 (overall reliability of predictors)               :  0.0007818689 

Testa0|b1(incorrect calibration given correct refinement)  :  0.01399315 

Testb1|a (refinement given correct calibration)            :  0.004033571 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.5634618  (sd 0.19191) 

NULL Deviance                : 0.056371 (sd 0.0088872)

Fit Deviance                 : 0.030607 (sd 0.020319)

Explained Deviance           : 0.025764 (sd 0.020648)

Percent Deviance Explained   : 45.465 (sd 36.975)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.4395 (sd 0.022042) 

Confusion Matrix: 

         observed

predicted    1    0

        1   10    6

        0   12 4974

AUC                          :  0.99461  (sd 0.0051384) 

Percent Correctly Classified :  99.64024  (sd 0.18344) 

Sensitivity                  :  0.4833333  (sd 0.41164) 

Specificity                  :  0.9987952  (sd 0.001404) 

Kappa                        :  0.4522786  (sd 0.35464) 

True Skill Statistic         :  0.4821285  (sd 0.41086) 

   Calibration Statistics 

Intercept (general calibration)                            :  1.372115  (sd 3.5484) 

Slope   (direction and variation in fit)                   :  4.221871  (sd 7.1765) 

Testa0b1 (overall reliability of predictors)               :  0.4119389  (sd 0.37086) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.4535522  (sd 0.30439) 

Testb1|a (refinement given correct calibration)            :  0.3517502  (sd 0.34396) 

Total time = 2.53 min