Species 015: Abies concolor

General Linear Models

Boosted Regression Trees

Generalized Linear Model Results

n(pres)                 =      374

n(abs)                  =      4636

n covariates considered =      9

total time for model fitting = 5.7min

Settings:

model family          : binomial

simplification method : AIC

Results:

number covariates in final model   : 38

Call:

glm(formula = response ~ I(bio05^2) + bio05 + bio15 + bio08 + 

    bio02 + I(bio02^2) + I(bio18^2) + bio12 + I(bio12^2) + bio18 + 

    I(bio15^2) + bio09 + I(bio09^2) + I(bio08^2) + di_all_model + 

    I(di_all_model^2) + bio15:bio08 + bio08:bio02 + bio15:bio12 + 

    bio08:bio12 + bio05:bio12 + bio05:bio15 + bio12:bio18 + bio02:bio18 + 

    bio05:bio02 + bio02:bio12 + bio15:bio18 + bio05:bio18 + bio08:bio18 + 

    bio18:bio09 + bio02:di_all_model + bio08:di_all_model + bio18:di_all_model + 

    bio15:di_all_model + bio05:bio09 + bio02:bio09 + bio05:bio08 + 

    bio12:di_all_model, family = model.family, data = dat, weights = weight, 

    na.action = "na.exclude")

Deviance Residuals: 

    Min       1Q   Median       3Q      Max  

-2.9081  -0.1226  -0.0143  -0.0002   3.2732  

Coefficients:

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

(Intercept)        -9.920e+01  1.791e+01  -5.539 3.04e-08 ***
I(bio05^2)         -1.827e-03  2.604e-04  -7.015 2.29e-12 ***

bio05               4.553e-01  1.005e-01   4.528 5.95e-06 ***

bio15               4.342e-01  1.065e-01   4.078 4.54e-05 ***

bio08               8.172e-02  3.879e-02   2.107 0.035140 *  

bio02               3.767e-01  1.468e-01   2.565 0.010304 *  

I(bio02^2)         -3.810e-03  6.695e-04  -5.691 1.26e-08 ***

I(bio18^2)          3.147e-04  9.061e-05   3.473 0.000515 ***

bio12               2.182e-02  7.992e-03   2.730 0.006332 ** 

I(bio12^2)         -3.878e-06  1.787e-06  -2.171 0.029959 *  

bio18              -4.286e-02  6.875e-02  -0.623 0.533048    

I(bio15^2)         -1.230e-03  4.438e-04  -2.771 0.005584 ** 

bio09              -9.847e-03  4.325e-02  -0.228 0.819899    

I(bio09^2)         -2.292e-04  5.080e-05  -4.512 6.42e-06 ***

I(bio08^2)         -1.020e-04  4.348e-05  -2.345 0.019037 *  

di_all_model       -2.022e-01  9.458e-02  -2.138 0.032498 *  

I(di_all_model^2)  -3.791e-04  2.273e-04  -1.668 0.095339 .  

bio15:bio08        -3.683e-04  1.931e-04  -1.907 0.056537 .  

bio08:bio02        -5.729e-04  1.855e-04  -3.089 0.002006 ** 

bio15:bio12        -1.310e-04  6.338e-05  -2.067 0.038688 *  

bio08:bio12        -1.085e-04  1.617e-05  -6.712 1.92e-11 ***

bio05:bio12         1.280e-04  2.314e-05   5.532 3.16e-08 ***

bio05:bio15        -9.084e-04  3.678e-04  -2.470 0.013510 *  

bio12:bio18        -1.095e-04  2.456e-05  -4.457 8.33e-06 ***

bio02:bio18         1.449e-03  2.816e-04   5.144 2.69e-07 ***

bio05:bio02         3.147e-03  7.233e-04   4.351 1.35e-05 ***

bio02:bio12        -1.571e-04  4.126e-05  -3.808 0.000140 ***

bio15:bio18        -8.559e-04  2.294e-04  -3.731 0.000191 ***

bio05:bio18        -7.813e-04  2.545e-04  -3.069 0.002145 ** 

bio08:bio18         3.450e-04  1.073e-04   3.215 0.001305 ** 

bio18:bio09         1.700e-04  5.876e-05   2.893 0.003822 ** 

bio02:di_all_model  1.114e-03  5.661e-04   1.967 0.049163 *  

bio08:di_all_model -4.773e-04  1.295e-04  -3.686 0.000227 ***

bio18:di_all_model  3.943e-04  1.547e-04   2.549 0.010792 *  

bio15:di_all_model  1.278e-03  4.498e-04   2.842 0.004482 ** 

bio05:bio09         4.817e-04  1.672e-04   2.881 0.003963 ** 

bio02:bio09        -6.912e-04  2.793e-04  -2.475 0.013340 *  

bio05:bio08         1.817e-04  1.212e-04   1.499 0.133834    

bio12:di_all_model -6.186e-05  3.077e-05  -2.010 0.044396 *  

---

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

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 2660.37  on 5009  degrees of freedom

Residual deviance:  989.27  on 4971  degrees of freedom

AIC: 1067.3

Number of Fisher Scoring iterations: 10

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.753489 

NULL Deviance                : 0.53101

Fit Deviance                 : 0.19746

Explained Deviance           : 0.33355

Percent Deviance Explained   : 62.814

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.11 

Confusion Matrix: 

         observed

predicted    1    0

        1  342  368

        0   32 4268

AUC                          :  0.9748 

Percent Correctly Classified :  92.01597 

Sensitivity                  :  0.9144385 

Specificity                  :  0.9206212 

Kappa                        :  0.5910004 

True Skill Statistic         :  0.8350597 

   Calibration Statistics 

Intercept (general calibration)                            :  0.0003561888 

Slope   (direction and variation in fit)                   :  1.000683 

Testa0b1 (overall reliability of predictors)               :  0.9998936 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9966926 

Testb1|a (refinement given correct calibration)            :  0.9888378 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.7258847  (sd 0.047021) 

NULL Deviance                : 0.531 (sd 0.0049635)

Fit Deviance                 : 0.22326 (sd 0.035157)

Explained Deviance           : 0.30774 (sd 0.03471)

Percent Deviance Explained   : 57.961 (sd 6.5564)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.103 (sd 0.0048305) 

Confusion Matrix: 

         observed

predicted    1    0

        1  330  412

        0   44 4224

AUC                          :  0.96584  (sd 0.011801) 

Percent Correctly Classified :  90.89766  (sd 1.5083) 

Sensitivity                  :  0.8825036  (sd 0.033353) 

Specificity                  :  0.9111337  (sd 0.017004) 

Kappa                        :  0.5489875  (sd 0.047837) 

True Skill Statistic         :  0.7936372  (sd 0.032104) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.08111887  (sd 0.29113) 

Slope   (direction and variation in fit)                   :  0.9163103  (sd 0.12839) 

Testa0b1 (overall reliability of predictors)               :  0.4292463  (sd 0.30007) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.4055091  (sd 0.22577) 

Testb1|a (refinement given correct calibration)            :  0.5723386  (sd 0.41606) 

Total time = 48.91 min

Boosted Regression Tree Modeling Results

Data:

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

n(pres)                 =      374

n(abs)                  =      4636

n covariates considered =      9

total time for model fitting = 0.73min

Settings:

random seed used             : 19680828

tree complexity              : 9

learning rate                : 0.0738

n(trees)                     : 300

model simplification         : cross-validation

n folds                      : 3

n covariates in final model  : 5

Relative influence of predictors in final model:

   Var  rel.inf

 bio12 27.91475

 bio08 23.23542

 bio05 18.26510

 bio04 15.66607

 bio15 14.91866

Important interactions in final model:

 v1 name1 v2 name2

  3 bio08  2 bio05

  4 bio12  2 bio05

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.9440413 

NULL Deviance                : 0.53101

Fit Deviance                 : 0.071447

Explained Deviance           : 0.45956

Percent Deviance Explained   : 86.545

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.23 

Confusion Matrix: 

         observed

predicted    1    0

        1  369   74

        0    5 4562

AUC                          :  0.9991 

Percent Correctly Classified :  98.42315 

Sensitivity                  :  0.986631 

Specificity                  :  0.984038 

Kappa                        :  0.8947872 

True Skill Statistic         :  0.970669 

   Calibration Statistics 

Intercept (general calibration)                            :  0.936978 

Slope   (direction and variation in fit)                   :  2.708824 

Testa0b1 (overall reliability of predictors)               :  0 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9719894 

Testb1|a (refinement given correct calibration)            :  0 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.713849  (sd 0.072171) 

NULL Deviance                : 0.531 (sd 0.0049635)

Fit Deviance                 : 0.23492 (sd 0.056383)

Explained Deviance           : 0.29608 (sd 0.056962)

Percent Deviance Explained   : 55.749 (sd 10.698)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.226 (sd 0.016465) 

Confusion Matrix: 

         observed

predicted    1    0

        1  291  180

        0   83 4456

AUC                          :  0.96821  (sd 0.015229) 

Percent Correctly Classified :  94.75022  (sd 0.98341) 

Sensitivity                  :  0.7783784  (sd 0.069545) 

Specificity                  :  0.961174  (sd 0.01089) 

Kappa                        :  0.6615237  (sd 0.052557) 

True Skill Statistic         :  0.7395524  (sd 0.067205) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.07511296  (sd 0.36254) 

Slope   (direction and variation in fit)                   :  0.8283396  (sd 0.13247) 

Testa0b1 (overall reliability of predictors)               :  0.2594719  (sd 0.33354) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.4020512  (sd 0.28796) 

Testb1|a (refinement given correct calibration)            :  0.3345779  (sd 0.35705) 

Total time = 13.83 min

MARS

Random Forests

MARS Model Results

Data:

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

n(pres)                 =      374

n(abs)                  =      4636

n covariates considered =      9

total time for model fitting = 0.1min

Settings:

random seed used             : 123

mars degree                  : 1

mars penalty                 : 2

Summary of Model:

      nsubsets   gcv    rss

bio12       15 100.0  100.0

bio15       12  64.8   65.4

bio04       11  60.4   60.9

bio05       10  57.6   58.1

bio02        5  24.2   25.2

bio18        4  20.8   21.8

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.7070328 

NULL Deviance                : 0.53101

Fit Deviance                 : 0.23329

Explained Deviance           : 0.29772

Percent Deviance Explained   : 56.066

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.09 

Confusion Matrix: 

         observed

predicted    1    0

        1  328  533

        0   46 4103

AUC                          :  0.9617 

Percent Correctly Classified :  88.44311 

Sensitivity                  :  0.8770053 

Specificity                  :  0.8850302 

Kappa                        :  0.4767055 

True Skill Statistic         :  0.7620355 

   Calibration Statistics 

Intercept (general calibration)                            :  0.0003802926 

Slope   (direction and variation in fit)                   :  1.000582 

Testa0b1 (overall reliability of predictors)               :  0.999915 

Testa0|b1(incorrect calibration given correct refinement)  :  0.996948 

Testb1|a (refinement given correct calibration)            :  0.9900516 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6892213  (sd 0.070387) 

NULL Deviance                : 0.531 (sd 0.0049635)

Fit Deviance                 : 0.2573 (sd 0.068444)

Explained Deviance           : 0.2737 (sd 0.067316)

Percent Deviance Explained   : 51.569 (sd 12.726)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.092 (sd 0.0042164) 

Confusion Matrix: 

         observed

predicted    1    0

        1  324  517

        0   50 4119

AUC                          :  0.95639  (sd 0.018567) 

Percent Correctly Classified :  88.68119  (sd 1.7011) 

Sensitivity                  :  0.866074  (sd 0.069866) 

Specificity                  :  0.8884692  (sd 0.016361) 

Kappa                        :  0.4811356  (sd 0.056328) 

True Skill Statistic         :  0.7545432  (sd 0.075063) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.02680593  (sd 0.31762) 

Slope   (direction and variation in fit)                   :  0.988848  (sd 0.2089) 

Testa0b1 (overall reliability of predictors)               :  0.3397809  (sd 0.19781) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.517833  (sd 0.27548) 

Testb1|a (refinement given correct calibration)            :  0.3449844  (sd 0.33271) 

Total time = 2.26 min

Random Forest Modeling Results

Data:

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

n(pres)                 =      374

n(abs)                  =      4636

n covariates considered =      9

total time for model fitting = 0.3min

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

bio12        31.4581 92.2498              66.3007          71.1686

bio08        37.1722 59.2817              57.6834          59.1066

bio05        39.6176 50.9003              55.0889          55.2130

bio04        30.6171 49.6858              49.8989          62.0867

bio09        33.9993 35.3315              47.9689          35.0978

bio15        35.7962 38.4387              47.4891          44.3776

bio18        32.9148 44.5145              46.0944          47.1211

bio02        38.8379 31.1848              45.9173          39.3897

di_all_model 22.7922 27.1159              33.3777          23.9322

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.7671461 

NULL Deviance                : 0.53101

Fit Deviance                 : 0.19501

Explained Deviance           : 0.336

Percent Deviance Explained   : 63.275

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.12 

Confusion Matrix: 

         observed

predicted    1    0

        1  343  358

        0   31 4278

AUC                          :  0.9769 

Percent Correctly Classified :  92.23553 

Sensitivity                  :  0.9171123 

Specificity                  :  0.9227783 

Kappa                        :  0.5991095 

True Skill Statistic         :  0.8398906 

   Calibration Statistics 

Intercept (general calibration)                            :  0.1855548 

Slope   (direction and variation in fit)                   :  1.194594 

Testa0b1 (overall reliability of predictors)               :  0.0003614805 

Testa0|b1(incorrect calibration given correct refinement)  :  0.8540328 

Testb1|a (refinement given correct calibration)            :  6.978172e-05 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.7668042  (sd 0.060553) 

NULL Deviance                : 0.531 (sd 0.0049635)

Fit Deviance                 : 0.1951 (sd 0.035361)

Explained Deviance           : 0.3359 (sd 0.035298)

Percent Deviance Explained   : 63.26 (sd 6.6545)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.4915 (sd 0.0088349) 

Confusion Matrix: 

         observed

predicted    1    0

        1  239   59

        0  135 4577

AUC                          :  0.97583  (sd 0.0099681) 

Percent Correctly Classified :  96.12739  (sd 1.0204) 

Sensitivity                  :  0.6391181  (sd 0.080042) 

Specificity                  :  0.987269  (sd 0.0076546) 

Kappa                        :  0.6907709  (sd 0.081793) 

True Skill Statistic         :  0.6263871  (sd 0.082078) 

   Calibration Statistics 

Intercept (general calibration)                            :  0.2179234  (sd 0.39086) 

Slope   (direction and variation in fit)                   :  1.199812  (sd 0.19587) 

Testa0b1 (overall reliability of predictors)               :  0.388223  (sd 0.34367) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.5054063  (sd 0.25855) 

Testb1|a (refinement given correct calibration)            :  0.3149526  (sd 0.32356) 

Total time = 7.71 min