Species 110: Pinus echinata

General Linear Models

Boosted Regression Trees

Generalized Linear Model Results

Data:

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

n(pres)                 =      891

n(abs)                  =      4312

n covariates considered =      7

total time for model fitting = 0.93min

Settings:

model family          : binomial

simplification method : AIC

Results:

number covariates in final model   : 16

Call:

glm(formula = response ~ bio02 + I(bio15^2) + I(di_all_model^2) + 

    bio11 + I(bio11^2) + bio08 + bio09 + I(bio09^2) + I(bio08^2) + 

    bio15 + I(bio02^2) + bio08:bio09 + bio11:bio08 + bio08:bio15 + 

    bio11:bio09 + bio09:bio15, family = model.family, data = dat, 

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

Deviance Residuals: 

    Min       1Q   Median       3Q      Max  

-2.2315  -0.4619  -0.1515  -0.0007   4.0032  

Coefficients:

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

(Intercept)       -5.083e+01  1.397e+01  -3.638 0.000275 ***

bio02              7.161e-01  2.140e-01   3.346 0.000820 ***

I(bio15^2)        -4.141e-03  1.744e-03  -2.374 0.017589 *  

I(di_all_model^2) -3.798e-04  3.987e-05  -9.527  < 2e-16 ***

bio11              1.747e-01  1.377e-02  12.683  < 2e-16 ***

I(bio11^2)        -8.507e-04  9.423e-05  -9.028  < 2e-16 ***

bio08             -4.897e-02  7.532e-03  -6.502 7.93e-11 ***

bio09             -5.088e-02  6.057e-03  -8.400  < 2e-16 ***

I(bio09^2)         1.076e-04  2.027e-05   5.309 1.10e-07 ***

I(bio08^2)         1.979e-04  2.258e-05   8.765  < 2e-16 ***

bio15              5.335e-01  7.416e-02   7.194 6.28e-13 ***

I(bio02^2)        -2.532e-03  8.219e-04  -3.081 0.002065 ** 

bio08:bio09        2.043e-04  2.115e-05   9.659  < 2e-16 ***

bio11:bio08       -2.076e-04  5.395e-05  -3.848 0.000119 ***

bio08:bio15       -2.385e-03  2.674e-04  -8.920  < 2e-16 ***

bio11:bio09       -2.318e-04  6.412e-05  -3.616 0.000299 ***

bio09:bio15       -4.252e-04  1.600e-04  -2.658 0.007858 ** 

---

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

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 4764.5  on 5202  degrees of freedom

Residual deviance: 2966.1  on 5186  degrees of freedom

AIC: 3000.1

Number of Fisher Scoring iterations: 9

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.6050781 

NULL Deviance                : 0.91572

Fit Deviance                 : 0.57007

Explained Deviance           : 0.34564

Percent Deviance Explained   : 37.746

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.24 

Confusion Matrix: 

         observed

predicted    1    0

        1  735  747

        0  156 3565

AUC                          :  0.9016 

Percent Correctly Classified :  82.64463 

Sensitivity                  :  0.8249158 

Specificity                  :  0.8267625 

Kappa                        :  0.5159274 

True Skill Statistic         :  0.6516783 

   Calibration Statistics 

Intercept (general calibration)                            :  9.241157e-05 

Slope   (direction and variation in fit)                   :  1.000284 

Testa0b1 (overall reliability of predictors)               :  0.999968 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9981004 

Testb1|a (refinement given correct calibration)            :  0.9939057 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.5991412  (sd 0.030324) 

NULL Deviance                : 0.91572 (sd 0.0042226)

Fit Deviance                 : 0.57844 (sd 0.034561)

Explained Deviance           : 0.33728 (sd 0.033923)

Percent Deviance Explained   : 36.834 (sd 3.7251)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.24 (sd 0) 

Confusion Matrix: 

         observed

predicted    1    0

        1  730  755

        0  161 3557

AUC                          :  0.89875  (sd 0.013171) 

Percent Correctly Classified :  82.39652  (sd 1.1645) 

Sensitivity                  :  0.8192884  (sd 0.035391) 

Specificity                  :  0.8249237  (sd 0.012235) 

Kappa                        :  0.5095933  (sd 0.029707) 

True Skill Statistic         :  0.6442121  (sd 0.037101) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.01721622  (sd 0.098573) 

Slope   (direction and variation in fit)                   :  0.9837447  (sd 0.13273) 

Testa0b1 (overall reliability of predictors)               :  0.5354908  (sd 0.2246) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.7076174  (sd 0.13975) 

Testb1|a (refinement given correct calibration)            :  0.3406387  (sd 0.23032) 

Total time = 21.47 min

Boosted Regression Tree Modeling Results

Data:

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

n(pres)                 =      891

n(abs)                  =      4312

n covariates considered =      7

total time for model fitting = 0.83min

Settings:

random seed used             : 19680828

tree complexity              : 9

learning rate                : 0.0657

n(trees)                     : 300

model simplification         : cross-validation

n folds                      : 3

n covariates in final model  : 6

Relative influence of predictors in final model:

          Var  rel.inf

        bio11 22.80530

        bio09 18.07084

        bio15 17.43762

        bio02 16.68363

 di_all_model 14.09921

        bio18 10.90339

Important interactions in final model:

 v1        name1 v2 name2

  6 di_all_model  4 bio15

  3        bio11  2 bio09

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.8292138 

NULL Deviance                : 0.91572

Fit Deviance                 : 0.32013

Explained Deviance           : 0.59559

Percent Deviance Explained   : 65.04

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.26 

Confusion Matrix: 

         observed

predicted    1    0

        1  824  334

        0   67 3978

AUC                          :  0.9766 

Percent Correctly Classified :  92.29291 

Sensitivity                  :  0.9248036 

Specificity                  :  0.9225417 

Kappa                        :  0.7573214 

True Skill Statistic         :  0.8473453 

   Calibration Statistics 

Intercept (general calibration)                            :  0.2957373 

Slope   (direction and variation in fit)                   :  1.566342 

Testa0b1 (overall reliability of predictors)               :  0 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9635748 

Testb1|a (refinement given correct calibration)            :  0 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6369129  (sd 0.044896) 

NULL Deviance                : 0.91572 (sd 0.0042226)

Fit Deviance                 : 0.52946 (sd 0.044996)

Explained Deviance           : 0.38627 (sd 0.045209)

Percent Deviance Explained   : 42.181 (sd 4.9182)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.26 (sd 0.0066667) 

Confusion Matrix: 

         observed

predicted    1    0

        1  696  532

        0  195 3780

AUC                          :  0.91708  (sd 0.017291) 

Percent Correctly Classified :  86.03004  (sd 2.1678) 

Sensitivity                  :  0.7811735  (sd 0.058183) 

Specificity                  :  0.8766449  (sd 0.015198) 

Kappa                        :  0.5722756  (sd 0.064911) 

True Skill Statistic         :  0.6578184  (sd 0.071526) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.02876412  (sd 0.1016) 

Slope   (direction and variation in fit)                   :  0.9230417  (sd 0.071315) 

Testa0b1 (overall reliability of predictors)               :  0.5313581  (sd 0.25142) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.612493  (sd 0.15169) 

Testb1|a (refinement given correct calibration)            :  0.3827561  (sd 0.24512) 

Total time = 12.04 min

MARS

Random Forests

MARS Model Results

Data:

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

n(pres)                 =      891

n(abs)                  =      4312

n covariates considered =      7

total time for model fitting = 0.12min

Settings:

random seed used             : 123

mars degree                  : 1

mars penalty                 : 2

Summary of Model:

             nsubsets   gcv    rss

bio08              15 100.0  100.0

bio02              13  78.4   78.8

di_all_model       13  66.6   67.4

bio09              11  58.7   59.5

bio18              11  58.7   59.5

bio11              11  57.8   58.6

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.5989253 

NULL Deviance                : 0.91572

Fit Deviance                 : 0.57775

Explained Deviance           : 0.33797

Percent Deviance Explained   : 36.908

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.2 

Confusion Matrix: 

         observed

predicted    1    0

        1  722  795

        0  169 3517

AUC                          :  0.8969 

Percent Correctly Classified :  81.47223 

Sensitivity                  :  0.8103255 

Specificity                  :  0.8156308 

Kappa                        :  0.4895245 

True Skill Statistic         :  0.6259563 

   Calibration Statistics 

Intercept (general calibration)                            :  -8.707437e-06 

Slope   (direction and variation in fit)                   :  1.000122 

Testa0b1 (overall reliability of predictors)               :  0.9999907 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9980898 

Testb1|a (refinement given correct calibration)            :  0.9971446 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.5995638  (sd 0.027376) 

NULL Deviance                : 0.91572 (sd 0.0042226)

Fit Deviance                 : 0.57366 (sd 0.019148)

Explained Deviance           : 0.34206 (sd 0.019696)

Percent Deviance Explained   : 37.353 (sd 2.1142)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.202 (sd 0.0078881) 

Confusion Matrix: 

         observed

predicted    1    0

        1  719  787

        0  172 3525

AUC                          :  0.89889  (sd 0.0078711) 

Percent Correctly Classified :  81.56984  (sd 1.1509) 

Sensitivity                  :  0.8069164  (sd 0.031025) 

Specificity                  :  0.8174875  (sd 0.014517) 

Kappa                        :  0.4904203  (sd 0.025037) 

True Skill Statistic         :  0.6244039  (sd 0.029771) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.02304299  (sd 0.082508) 

Slope   (direction and variation in fit)                   :  0.9739517  (sd 0.050136) 

Testa0b1 (overall reliability of predictors)               :  0.7804725  (sd 0.15225) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.6814571  (sd 0.21341) 

Testb1|a (refinement given correct calibration)            :  0.6793173  (sd 0.20537) 

Total time = 2.15 min

Random Forest Modeling Results

Data:

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

n(pres)                 =      891

n(abs)                  =      4312

n covariates considered =      7

total time for model fitting = 0.35min

Settings:

random seed used                       : 19680828

n covariates considered at each split  : 3

n trees                                : 1000

Relative performance of predictors in final model:

                   0        1 MeanDecreaseAccuracy MeanDecreaseGini

di_all_model 46.6217 105.1240             102.3158         126.6017

bio09        72.7204  53.8646              97.8261         140.8348

bio02        42.7333  86.7558              97.2139         133.4032

bio15        61.4781  64.6374              96.6075         107.3622

bio08        60.5527  74.9905              91.9397         165.7882

bio11        55.9656  65.5638              82.3063         137.3632

bio18        52.3186  50.3539              79.6719         121.0719

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.6986533 

NULL Deviance                : 0.91572

Fit Deviance                 : 0.46617

Explained Deviance           : 0.44955

Percent Deviance Explained   : 49.093

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.22 

Confusion Matrix: 

         observed

predicted    1    0

        1  780  537

        0  111 3775

AUC                          :  0.9389 

Percent Correctly Classified :  87.54565 

Sensitivity                  :  0.8754209 

Specificity                  :  0.8754638 

Kappa                        :  0.6311758 

True Skill Statistic         :  0.7508847 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.04086372 

Slope   (direction and variation in fit)                   :  0.9434321 

Testa0b1 (overall reliability of predictors)               :  0.205927 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9125061 

Testb1|a (refinement given correct calibration)            :  0.07600174 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6983316  (sd 0.020369) 

NULL Deviance                : 0.91572 (sd 0.0042226)

Fit Deviance                 : 0.46948 (sd 0.032756)

Explained Deviance           : 0.44624 (sd 0.031427)

Percent Deviance Explained   : 48.736 (sd 3.4849)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.503 (sd 0.017192) 

Confusion Matrix: 

         observed

predicted    1    0

        1  562  211

        0  329 4101

AUC                          :  0.93733  (sd 0.010433) 

Percent Correctly Classified :  89.62318  (sd 1.0211) 

Sensitivity                  :  0.6306617  (sd 0.050937) 

Specificity                  :  0.9510479  (sd 0.0106) 

Kappa                        :  0.6136437  (sd 0.040717) 

True Skill Statistic         :  0.5817095  (sd 0.048801) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.01190836  (sd 0.15751) 

Slope   (direction and variation in fit)                   :  0.9776165  (sd 0.12183) 

Testa0b1 (overall reliability of predictors)               :  0.526944  (sd 0.30342) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.570928  (sd 0.18754) 

Testb1|a (refinement given correct calibration)            :  0.4995247  (sd 0.35714) 

Total time = 9.52 min