Species 121: Pinus palustris

General Linear Models

Boosted Regression Trees

Generalized Linear Model Results

Data:

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

n(pres)                 =      596

n(abs)                  =      4602

n covariates considered =      9

total time for model fitting = 1.67min

Settings:

model family          : binomial

simplification method : AIC

Results:

number covariates in final model   : 30

Call:

glm(formula = response ~ bio18 + I(di_all_model^2) + bio10 + 

    I(bio10^2) + bio02 + di_all_model + I(bio02^2) + I(bio09^2) + 

    I(bio15^2) + bio12 + bio08 + I(bio08^2) + bio09 + I(bio12^2) + 

    bio03 + bio18:bio02 + bio10:di_all_model + di_all_model:bio12 + 

    di_all_model:bio08 + bio08:bio09 + bio18:bio08 + bio12:bio08 + 

    bio10:bio12 + bio18:bio09 + bio12:bio09 + bio18:di_all_model + 

    bio18:bio12 + bio08:bio03 + bio10:bio03 + bio02:bio09, family = model.family, 

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

Deviance Residuals: 

    Min       1Q   Median       3Q      Max  

-2.3420  -0.4065  -0.1164  -0.0126   3.2121  

Coefficients:

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

(Intercept)        -6.550e+02  8.407e+01  -7.792 6.61e-15 ***

bio18               1.899e-01  3.212e-02   5.914 3.35e-09 ***

I(di_all_model^2)   3.708e-04  1.207e-04   3.072 0.002123 ** 

bio10               3.502e+00  5.662e-01   6.186 6.19e-10 ***

I(bio10^2)         -4.827e-03  1.084e-03  -4.455 8.40e-06 ***

bio02               1.006e+00  1.999e-01   5.032 4.85e-07 ***

di_all_model       -3.627e-01  1.294e-01  -2.803 0.005070 ** 

I(bio02^2)         -3.745e-03  6.935e-04  -5.400 6.66e-08 ***

I(bio09^2)          2.524e-04  6.306e-05   4.002 6.28e-05 ***

I(bio15^2)         -1.757e-03  3.757e-04  -4.676 2.93e-06 ***

bio12               3.074e-02  2.378e-02   1.292 0.196204    

bio08              -1.176e-01  3.255e-02  -3.614 0.000302 ***

I(bio08^2)          2.808e-04  5.386e-05   5.214 1.85e-07 ***

bio09              -9.421e-02  6.129e-02  -1.537 0.124292    

I(bio12^2)          1.408e-05  3.842e-06   3.665 0.000247 ***

bio03               4.486e+00  1.505e+00   2.980 0.002878 ** 

bio18:bio02        -5.341e-04  1.393e-04  -3.833 0.000127 ***

bio10:di_all_model  1.421e-03  5.197e-04   2.733 0.006267 ** 

di_all_model:bio12 -6.617e-05  2.177e-05  -3.040 0.002369 ** 

di_all_model:bio08 -2.031e-04  6.360e-05  -3.194 0.001406 ** 

bio08:bio09        -3.667e-04  5.551e-05  -6.607 3.93e-11 ***

bio18:bio08        -3.005e-04  4.810e-05  -6.247 4.18e-10 ***

bio12:bio08         6.292e-05  1.265e-05   4.972 6.61e-07 ***

bio10:bio12        -1.808e-04  8.625e-05  -2.097 0.036038 *  

bio18:bio09         2.038e-04  5.219e-05   3.905 9.43e-05 ***

bio12:bio09        -8.138e-05  2.649e-05  -3.072 0.002124 ** 

bio18:di_all_model  1.170e-04  5.447e-05   2.149 0.031668 *  

bio18:bio12        -5.110e-05  1.223e-05  -4.178 2.94e-05 ***

bio08:bio03         2.809e-03  7.686e-04   3.655 0.000257 ***

bio10:bio03        -1.893e-02  5.899e-03  -3.209 0.001332 ** 

bio02:bio09         9.347e-04  3.507e-04   2.666 0.007686 ** 

---

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

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 3702.5  on 5197  degrees of freedom

Residual deviance: 2343.8  on 5167  degrees of freedom

AIC: 2405.8

Number of Fisher Scoring iterations: 10

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.5474607 

NULL Deviance                : 0.7123

Fit Deviance                 : 0.4509

Explained Deviance           : 0.26139

Percent Deviance Explained   : 36.698

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.17 

Confusion Matrix: 

         observed

predicted    1    0

        1  490  850

        0  106 3752

AUC                          :  0.9013 

Percent Correctly Classified :  81.60831 

Sensitivity                  :  0.8221477 

Specificity                  :  0.8152977 

Kappa                        :  0.4130333 

True Skill Statistic         :  0.6374453 

   Calibration Statistics 

Intercept (general calibration)                            :  0.0002058589 

Slope   (direction and variation in fit)                   :  1.000318 

Testa0b1 (overall reliability of predictors)               :  0.9999701 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9978341 

Testb1|a (refinement given correct calibration)            :  0.9942275 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.5273322  (sd 0.042227) 

NULL Deviance                : 0.7123 (sd 0.0050141)

Fit Deviance                 : 0.46676 (sd 0.028679)

Explained Deviance           : 0.24554 (sd 0.025528)

Percent Deviance Explained   : 34.486 (sd 3.7356)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.174 (sd 0.005164) 

Confusion Matrix: 

         observed

predicted    1    0

        1  472  849

        0  124 3753

AUC                          :  0.89475  (sd 0.013319) 

Percent Correctly Classified :  81.27798  (sd 2.2914) 

Sensitivity                  :  0.7919774  (sd 0.056255) 

Specificity                  :  0.8154683  (sd 0.028277) 

Kappa                        :  0.3984832  (sd 0.037797) 

True Skill Statistic         :  0.6074457  (sd 0.050529) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.06126048  (sd 0.14341) 

Slope   (direction and variation in fit)                   :  0.9489351  (sd 0.12383) 

Testa0b1 (overall reliability of predictors)               :  0.5602396  (sd 0.25489) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.5460184  (sd 0.20544) 

Testb1|a (refinement given correct calibration)            :  0.4462318  (sd 0.24402) 

Total time = 14.85 min

Boosted Regression Tree Modeling Results

Data:

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

n(pres)                 =      596

n(abs)                  =      4602

n covariates considered =      9

total time for model fitting = 0.33min

Settings:

random seed used             : 19680828

tree complexity              : 9

learning rate                : 0.1066

n(trees)                     : 300

model simplification         : cross-validation

n folds                      : 3

n covariates in final model  : 8

Relative influence of predictors in final model:

          Var   rel.inf

        bio18 24.259672

 di_all_model 14.734167

        bio12 14.083602

        bio09 11.914515

        bio03  9.613607

        bio10  8.960288

        bio15  8.588263

        bio08  7.845887

Important interactions in final model:

 v1 name1 v2 name2

  3 bio09  2 bio08

  7 bio18  3 bio09

  3 bio09  1 bio03

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.9053456 

NULL Deviance                : 0.7123

Fit Deviance                 : 0.16315

Explained Deviance           : 0.54915

Percent Deviance Explained   : 77.096

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.24 

Confusion Matrix: 

         observed

predicted    1    0

        1  574  152

        0   22 4450

AUC                          :  0.9956 

Percent Correctly Classified :  96.65256 

Sensitivity                  :  0.9630872 

Specificity                  :  0.9669709 

Kappa                        :  0.8494177 

True Skill Statistic         :  0.9300581 

   Calibration Statistics 

Intercept (general calibration)                            :  1.144748 

Slope   (direction and variation in fit)                   :  2.637982 

Testa0b1 (overall reliability of predictors)               :  0 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9908758 

Testb1|a (refinement given correct calibration)            :  0 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.635463  (sd 0.043123) 

NULL Deviance                : 0.7123 (sd 0.0050141)

Fit Deviance                 : 0.39466 (sd 0.040571)

Explained Deviance           : 0.31764 (sd 0.039167)

Percent Deviance Explained   : 44.603 (sd 5.5858)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.233 (sd 0.012517) 

Confusion Matrix: 

         observed

predicted    1    0

        1  424  360

        0  172 4242

AUC                          :  0.93237  (sd 0.016339) 

Percent Correctly Classified :  89.76206  (sd 1.5828) 

Sensitivity                  :  0.7115537  (sd 0.060453) 

Specificity                  :  0.9217496  (sd 0.015632) 

Kappa                        :  0.5575377  (sd 0.05666) 

True Skill Statistic         :  0.6333033  (sd 0.063237) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.02950947  (sd 0.21816) 

Slope   (direction and variation in fit)                   :  0.8922371  (sd 0.10992) 

Testa0b1 (overall reliability of predictors)               :  0.3507934  (sd 0.31549) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.5279953  (sd 0.27149) 

Testb1|a (refinement given correct calibration)            :  0.3207985  (sd 0.2894) 

Total time = 5.05 min

MARS

Random Forests

MARS Model Results

Data:

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

n(pres)                 =      596

n(abs)                  =      4602

n covariates considered =      9

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              13 100.0  100.0

di_all_model       12  73.3   74.2

bio10              11  66.6   67.6

bio12               9  48.2   49.7

bio03               8  41.0   42.6

bio02               7  37.0   38.6

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.5218758 

NULL Deviance                : 0.7123

Fit Deviance                 : 0.47613

Explained Deviance           : 0.23616

Percent Deviance Explained   : 33.155

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.14 

Confusion Matrix: 

         observed

predicted    1    0

        1  479  910

        0  117 3692

AUC                          :  0.8905 

Percent Correctly Classified :  80.2424 

Sensitivity                  :  0.8036913 

Specificity                  :  0.8022599 

Kappa                        :  0.3837295 

True Skill Statistic         :  0.6059512 

   Calibration Statistics 

Intercept (general calibration)                            :  8.988728e-05 

Slope   (direction and variation in fit)                   :  1.000187 

Testa0b1 (overall reliability of predictors)               :  0.9999856 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9978677 

Testb1|a (refinement given correct calibration)            :  0.9962917 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.5107448  (sd 0.048975) 

NULL Deviance                : 0.7123 (sd 0.0050141)

Fit Deviance                 : 0.49696 (sd 0.035185)

Explained Deviance           : 0.21534 (sd 0.035064)

Percent Deviance Explained   : 30.232 (sd 4.9355)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.136 (sd 0.0084327) 

Confusion Matrix: 

         observed

predicted    1    0

        1  473  913

        0  123 3689

AUC                          :  0.88749  (sd 0.015928) 

Percent Correctly Classified :  80.06643  (sd 1.8473) 

Sensitivity                  :  0.7936441  (sd 0.07503) 

Specificity                  :  0.8015862  (sd 0.02613) 

Kappa                        :  0.3775721  (sd 0.029524) 

True Skill Statistic         :  0.5952303  (sd 0.060692) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.0184341  (sd 0.21888) 

Slope   (direction and variation in fit)                   :  0.9784524  (sd 0.16983) 

Testa0b1 (overall reliability of predictors)               :  0.495111  (sd 0.32679) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.4820887  (sd 0.20811) 

Testb1|a (refinement given correct calibration)            :  0.46188  (sd 0.34208) 

Total time = 0.9 min

Random Forest Modeling Results

Data:

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

n(pres)                 =      596

n(abs)                  =      4602

n covariates considered =      9

total time for model fitting = 0.17min

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

di_all_model 88.6453 42.1698             100.7147          90.8743

bio18        59.2930 95.6917              88.0053         127.4902

bio12        62.7251 52.2161              85.9870          95.2534

bio09        72.0604 23.7511              85.2571          71.8921

bio10        57.4405 39.1643              71.4027          55.7006

bio15        57.1328 20.3051              67.2540          51.4904

bio02        49.7630 26.2897              57.2972          59.0714

bio03        41.3582 30.7754              55.3013          56.0602

bio08        22.7901 55.1817              32.6003          58.2635

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.6687144 

NULL Deviance                : 0.7123

Fit Deviance                 : 0.36041

Explained Deviance           : 0.35189

Percent Deviance Explained   : 49.402

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.15 

Confusion Matrix: 

         observed

predicted    1    0

        1  514  645

        0   82 3957

AUC                          :  0.9439 

Percent Correctly Classified :  86.01385 

Sensitivity                  :  0.8624161 

Specificity                  :  0.8598435 

Kappa                        :  0.5118248 

True Skill Statistic         :  0.7222597 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.04321277 

Slope   (direction and variation in fit)                   :  0.9867685 

Testa0b1 (overall reliability of predictors)               :  0.8150174 

Testa0|b1(incorrect calibration given correct refinement)  :  0.591938 

Testb1|a (refinement given correct calibration)            :  0.7271293 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.670632  (sd 0.034403) 

NULL Deviance                : 0.7123 (sd 0.0050141)

Fit Deviance                 : 0.35631 (sd 0.029731)

Explained Deviance           : 0.35598 (sd 0.027589)

Percent Deviance Explained   : 49.988 (sd 4.0149)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.459 (sd 0.020923) 

Confusion Matrix: 

         observed

predicted    1    0

        1  342  160

        0  254 4442

AUC                          :  0.94546  (sd 0.010032) 

Percent Correctly Classified :  92.03341  (sd 0.97172) 

Sensitivity                  :  0.574096  (sd 0.063537) 

Specificity                  :  0.9652314  (sd 0.0064767) 

Kappa                        :  0.5780545  (sd 0.054839) 

True Skill Statistic         :  0.5393275  (sd 0.06432) 

   Calibration Statistics 

Intercept (general calibration)                            :  0.0385748  (sd 0.14525) 

Slope   (direction and variation in fit)                   :  1.072897  (sd 0.12649) 

Testa0b1 (overall reliability of predictors)               :  0.5642242  (sd 0.23111) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.6256581  (sd 0.1999) 

Testb1|a (refinement given correct calibration)            :  0.4715543  (sd 0.30981) 

Total time = 3.38 min