Species 016: Abies fraseri

General Linear Models

Boosted Regression Trees

Generalized Linear Model Results

Data:

G:\Results01\16\Model\glm_1\CovariateCorrelationOutputMDS_initial.csv

n(pres)                 =      20

n(abs)                  =      4992

n covariates considered =      7

total time for model fitting = 0.4min

Settings:

model family          : binomial

simplification method : AIC

Results:

number covariates in final model   : 7

Call:

glm(formula = response ~ I(bio05^2) + bio02 + bio12 + I(bio18^2) + 

    I(bio08^2) + di_all_model + I(bio14^2), family = model.family, 

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

Deviance Residuals: 

    Min       1Q   Median       3Q      Max  

-1.1697  -0.0035  -0.0004   0.0000   3.2897  

Coefficients:

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

(Intercept)  -3.916e+01  1.135e+01  -3.450 0.000561 ***

I(bio05^2)   -4.967e-04  1.140e-04  -4.358 1.32e-05 ***

bio02         3.374e-01  1.054e-01   3.202 0.001364 ** 

bio12         3.824e-02  9.529e-03   4.012 6.01e-05 ***

I(bio18^2)   -1.348e-04  4.708e-05  -2.864 0.004178 ** 

I(bio08^2)    2.189e-04  6.458e-05   3.390 0.000699 ***

di_all_model -2.138e-01  4.467e-02  -4.786 1.70e-06 ***

I(bio14^2)   -5.007e-04  3.098e-04  -1.616 0.106019    

---

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

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 260.87  on 5011  degrees of freedom

Residual deviance: 112.77  on 5004  degrees of freedom

AIC: 128.77

Number of Fisher Scoring iterations: 13

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.4856269 

NULL Deviance                : 0.05205

Fit Deviance                 : 0.022501

Explained Deviance           : 0.029549

Percent Deviance Explained   : 56.771

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.01 

Confusion Matrix: 

         observed

predicted    1    0

        1   19  172

        0    1 4820

AUC                          :  0.9912 

Percent Correctly Classified :  96.54828 

Sensitivity                  :  0.95 

Specificity                  :  0.9655449 

Kappa                        :  0.1741284 

True Skill Statistic         :  0.9155449 

   Calibration Statistics 

Intercept (general calibration)                            :  0.01746305 

Slope   (direction and variation in fit)                   :  1.010514 

Testa0b1 (overall reliability of predictors)               :  0.9977419 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9898054 

Testb1|a (refinement given correct calibration)            :  0.947366 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.3445363  (sd 0.21997) 

NULL Deviance                : 0.05205 (sd 5.5597e-05)

Fit Deviance                 : 0.030423 (sd 0.011879)

Explained Deviance           : 0.021627 (sd 0.011851)

Percent Deviance Explained   : 41.562 (sd 22.763)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.012 (sd 0.0042164) 

Confusion Matrix: 

         observed

predicted    1    0

        1   18  200

        0    2 4792

AUC                          :  0.98468  (sd 0.010283) 

Percent Correctly Classified :  95.97092  (sd 1.5605) 

Sensitivity                  :  0.9  (sd 0.21082) 

Specificity                  :  0.9599491  (sd 0.016065) 

Kappa                        :  0.1571417  (sd 0.05982) 

True Skill Statistic         :  0.8599491  (sd 0.20341) 

   Calibration Statistics 

Intercept (general calibration)                            :  0.4806089  (sd 2.7546) 

Slope   (direction and variation in fit)                   :  1.286672  (sd 1.2606) 

Testa0b1 (overall reliability of predictors)               :  0.5682104  (sd 0.34) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.7122129  (sd 0.22192) 

Testb1|a (refinement given correct calibration)            :  0.4200923  (sd 0.31964) 

Total time = 6.3 min

Boosted Regression Tree Modeling Results

Data:

G:\Results01\16\Model\brt_1\CovariateCorrelationOutputMDS_initial.csv

n(pres)                 =      20

n(abs)                  =      4992

n covariates considered =      7

total time for model fitting = 0.44min

Settings:

random seed used             : 19680828

tree complexity              : 9

learning rate                : 0.0317

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 55.11439

 bio14 44.88561

Important interactions in final model:

 v1 name1 v2 name2

  2 bio14  1 bio05

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.9238353 

NULL Deviance                : 0.05205

Fit Deviance                 : 0.006486

Explained Deviance           : 0.045564

Percent Deviance Explained   : 87.539

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.09 

Confusion Matrix: 

         observed

predicted    1    0

        1   20   20

        0    0 4972

AUC                          :  0.9998 

Percent Correctly Classified :  99.60096 

Sensitivity                  :  1 

Specificity                  :  0.9959936 

Kappa                        :  0.6648837 

True Skill Statistic         :  0.9959936 

   Calibration Statistics 

Intercept (general calibration)                            :  1.675249 

Slope   (direction and variation in fit)                   :  2.666351 

Testa0b1 (overall reliability of predictors)               :  0.0004114937 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9740555 

Testb1|a (refinement given correct calibration)            :  7.86539e-05 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.119987  (sd 0.22836) 

NULL Deviance                : 0.05205 (sd 5.5597e-05)

Fit Deviance                 : 0.05338 (sd 0.019168)

Explained Deviance           : -0.00133 (sd 0.019171)

Percent Deviance Explained   : -2.5574 (sd 36.82)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.089 (sd 0.11083) 

Confusion Matrix: 

         observed

predicted    1    0

        1    5   63

        0   15 4929

AUC                          :  0.95513  (sd 0.075867) 

Percent Correctly Classified :  98.44451  (sd 0.95188) 

Sensitivity                  :  0.25  (sd 0.26352) 

Specificity                  :  0.987388  (sd 0.0098456) 

Kappa                        :  0.1191744  (sd 0.16753) 

True Skill Statistic         :  0.237388  (sd 0.26053) 

   Calibration Statistics 

Intercept (general calibration)                            :  -1.205048  (sd 1.5421) 

Slope   (direction and variation in fit)                   :  0.5844538  (sd 0.29831) 

Testa0b1 (overall reliability of predictors)               :  0.1231621  (sd 0.16041) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.2676612  (sd 0.19577) 

Testb1|a (refinement given correct calibration)            :  0.1238347  (sd 0.1905) 

Total time = 5.2 min

MARS

Random Forests

MARS Model Results

Data:

G:\Results01\16\Model\mars_1\CovariateCorrelationOutputMDS_initial.csv

n(pres)                 =      20

n(abs)                  =      4992

n covariates considered =      7

total time for model fitting = 0.1min

Settings:

random seed used             : 123

mars degree                  : 1

mars penalty                 : 2

Summary of Model:

             nsubsets   gcv    rss

bio05               8 100.0  100.0

bio12               8  88.2   88.7

bio14               8  88.2   88.7

di_all_model        3  16.6   20.7

bio08               1   5.6    9.3

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.4502662 

NULL Deviance                : 0.05205

Fit Deviance                 : 0.028718

Explained Deviance           : 0.023332

Percent Deviance Explained   : 44.826

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.01 

Confusion Matrix: 

         observed

predicted    1    0

        1   15  180

        0    5 4812

AUC                          :  0.9786 

Percent Correctly Classified :  96.30886 

Sensitivity                  :  0.75 

Specificity                  :  0.9639423 

Kappa                        :  0.133261 

True Skill Statistic         :  0.7139423 

   Calibration Statistics 

Intercept (general calibration)                            :  0.002699278 

Slope   (direction and variation in fit)                   :  1.001874 

Testa0b1 (overall reliability of predictors)               :  0.9997559 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9897411 

Testb1|a (refinement given correct calibration)            :  0.9856641 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.2777736  (sd 0.35647) 

NULL Deviance                : 0.05205 (sd 5.5597e-05)

Fit Deviance                 : 0.064123 (sd 0.06129)

Explained Deviance           : -0.012073 (sd 0.061277)

Percent Deviance Explained   : -23.167 (sd 117.7)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.01 (sd 0) 

Confusion Matrix: 

         observed

predicted    1    0

        1   11  191

        0    9 4801

AUC                          :  0.83927  (sd 0.22556) 

Percent Correctly Classified :  96.01001  (sd 1.0709) 

Sensitivity                  :  0.55  (sd 0.4378) 

Specificity                  :  0.9617443  (sd 0.010891) 

Kappa                        :  0.09097674  (sd 0.083287) 

True Skill Statistic         :  0.5117443  (sd 0.4362) 

   Calibration Statistics 

Intercept (general calibration)                            :  -2.447042  (sd 5.5385) 

Slope   (direction and variation in fit)                   :  1.588409  (sd 3.6173) 

Testa0b1 (overall reliability of predictors)               :  0.4277812  (sd 0.38558) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.5925641  (sd 0.24593) 

Testb1|a (refinement given correct calibration)            :  0.3344024  (sd 0.32192) 

Total time = 0.88 min

Random Forest Modeling Results

Data:

G:\Results01\16\Model\rf_1\CovariateCorrelationOutputMDS_initial.csv

n(pres)                 =      20

n(abs)                  =      4992

n covariates considered =      7

total time for model fitting = 0.14min

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

bio08        22.1651 -1.1370              22.9288           4.2575

bio05        17.0463 21.4918              18.5589           3.9701

bio12        13.3191 10.4016              14.6391           4.5877

bio18        12.4604 -5.9868              12.2227           3.1675

bio14         9.3997  4.0751              10.5854           3.6228

bio02         9.6192  1.1771              10.1151           2.6032

di_all_model -3.9746 13.2270               0.5199           3.0839

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.3189668 

NULL Deviance                : 0.05205

Fit Deviance                 : 0.034084

Explained Deviance           : 0.017966

Percent Deviance Explained   : 34.516

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.01 

Confusion Matrix: 

         observed

predicted    1    0

        1   17  173

        0    3 4819

AUC                          :  0.936 

Percent Correctly Classified :  96.48843 

Sensitivity                  :  0.85 

Specificity                  :  0.9653446 

Kappa                        :  0.1558091 

True Skill Statistic         :  0.8153446 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.3008491 

Slope   (direction and variation in fit)                   :  1.095711 

Testa0b1 (overall reliability of predictors)               :  0.0182118 

Testa0|b1(incorrect calibration given correct refinement)  :  0.007410161 

Testb1|a (refinement given correct calibration)            :  0.3592205 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.3213132  (sd 0.24411) 

NULL Deviance                : 0.05205 (sd 5.5597e-05)

Fit Deviance                 : 0.032148 (sd 0.012852)

Explained Deviance           : 0.019902 (sd 0.012835)

Percent Deviance Explained   : 38.244 (sd 24.651)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.4205 (sd 0.028524) 

Confusion Matrix: 

         observed

predicted    1    0

        1    1    8

        0   19 4984

AUC                          :  0.93698  (sd 0.11012) 

Percent Correctly Classified :  99.4614  (sd 0.26636) 

Sensitivity                  :  0.05  (sd 0.15811) 

Specificity                  :  0.9983984  (sd 0.0024588) 

Kappa                        :  0.06497629  (sd 0.21111) 

True Skill Statistic         :  0.0483984  (sd 0.15869) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.02095493  (sd 1.4321) 

Slope   (direction and variation in fit)                   :  1.226665  (sd 0.61931) 

Testa0b1 (overall reliability of predictors)               :  0.5330983  (sd 0.26014) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.7092371  (sd 0.2689) 

Testb1|a (refinement given correct calibration)            :  0.3739159  (sd 0.20802) 

Total time = 2.36 min