Species 022: Abies procera

General Linear Models

Boosted Regression Trees

Generalized Linear Model Results

Data:

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

n(pres)                 =      209

n(abs)                  =      2397

n covariates considered =      6

total time for model fitting = 0.18min

Settings:

model family          : binomial

simplification method : AIC

Results:

number covariates in final model   : 11

Call:

glm(formula = response ~ I(bio10^2) + I(bio06^2) + bio06 + bio17 + 

    I(bio17^2) + I(bio03^2) + bio15 + bio10 + I(bio15^2) + bio06:bio10 + 

    bio17:bio10, family = model.family, data = dat, weights = weight, 

    na.action = "na.exclude")

Deviance Residuals: 

    Min       1Q   Median       3Q      Max  

-2.2901  -0.0783  -0.0035   0.0000   4.3421  

Coefficients:

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

(Intercept) -2.277e+02  3.872e+01  -5.880 4.09e-09 ***

I(bio10^2)  -4.884e-03  1.067e-03  -4.580 4.65e-06 ***

I(bio06^2)  -3.964e-03  4.052e-04  -9.782  < 2e-16 ***

bio06       -1.096e+00  1.645e-01  -6.664 2.67e-11 ***

bio17        2.597e-01  5.432e-02   4.780 1.75e-06 ***

I(bio17^2)  -4.563e-04  8.158e-05  -5.593 2.23e-08 ***

I(bio03^2)   5.738e-03  8.948e-04   6.413 1.42e-10 ***

bio15        1.962e+00  5.954e-01   3.294 0.000986 ***

bio10        1.590e+00  3.562e-01   4.465 8.01e-06 ***

I(bio15^2)  -1.444e-02  4.846e-03  -2.980 0.002879 ** 

bio06:bio10  4.859e-03  1.011e-03   4.805 1.54e-06 ***

bio17:bio10 -5.621e-04  3.135e-04  -1.793 0.072970 .  

---

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

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 1455.5  on 2605  degrees of freedom

Residual deviance:  592.1  on 2594  degrees of freedom

AIC: 616.1

Number of Fisher Scoring iterations: 10

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.7344014 

NULL Deviance                : 0.55851

Fit Deviance                 : 0.22721

Explained Deviance           : 0.3313

Percent Deviance Explained   : 59.319

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.14 

Confusion Matrix: 

         observed

predicted    1    0

        1  190  215

        0   19 2182

AUC                          :  0.9704 

Percent Correctly Classified :  91.02072 

Sensitivity                  :  0.9090909 

Specificity                  :  0.9103045 

Kappa                        :  0.5738003 

True Skill Statistic         :  0.8193955 

   Calibration Statistics 

Intercept (general calibration)                            :  0.0006626956 

Slope   (direction and variation in fit)                   :  1.001492 

Testa0b1 (overall reliability of predictors)               :  0.9997753 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9978448 

Testb1|a (refinement given correct calibration)            :  0.9832256 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.7237628  (sd 0.072524) 

NULL Deviance                : 0.5585 (sd 0.0058812)

Fit Deviance                 : 0.24682 (sd 0.066187)

Explained Deviance           : 0.31168 (sd 0.064975)

Percent Deviance Explained   : 55.831 (sd 11.708)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.144 (sd 0.0069921) 

Confusion Matrix: 

         observed

predicted    1    0

        1  185  215

        0   24 2182

AUC                          :  0.96702  (sd 0.016913) 

Percent Correctly Classified :  90.82777  (sd 1.3085) 

Sensitivity                  :  0.8852381  (sd 0.075006) 

Specificity                  :  0.9102794  (sd 0.014285) 

Kappa                        :  0.5620862  (sd 0.048085) 

True Skill Statistic         :  0.7955175  (sd 0.072903) 

   Calibration Statistics 

Intercept (general calibration)                            :  0.05080208  (sd 0.454) 

Slope   (direction and variation in fit)                   :  1.083914  (sd 0.39097) 

Testa0b1 (overall reliability of predictors)               :  0.2694977  (sd 0.22233) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.540263  (sd 0.27155) 

Testb1|a (refinement given correct calibration)            :  0.2089948  (sd 0.22264) 

Total time = 2.25 min

Boosted Regression Tree Modeling Results

Data:

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

n(pres)                 =      209

n(abs)                  =      2397

n covariates considered =      6

total time for model fitting = 0.25min

Settings:

random seed used             : 19680828

tree complexity              : 6

learning rate                : 0.0399

n(trees)                     : 300

model simplification         : cross-validation

n folds                      : 3

n covariates in final model  : 3

Relative influence of predictors in final model:

   Var  rel.inf

 bio06 44.34037

 bio17 28.97895

 bio10 26.68068

Important interactions in final model:

 v1 name1 v2 name2

  2 bio10  1 bio06

  3 bio17  2 bio10

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.8352615 

NULL Deviance                : 0.55851

Fit Deviance                 : 0.16739

Explained Deviance           : 0.39112

Percent Deviance Explained   : 70.029

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.16 

Confusion Matrix: 

         observed

predicted    1    0

        1  197  148

        0   12 2249

AUC                          :  0.985 

Percent Correctly Classified :  93.86032 

Sensitivity                  :  0.9425837 

Specificity                  :  0.9382562 

Kappa                        :  0.6791416 

True Skill Statistic         :  0.8808399 

   Calibration Statistics 

Intercept (general calibration)                            :  0.496275 

Slope   (direction and variation in fit)                   :  1.557743 

Testa0b1 (overall reliability of predictors)               :  6.254758e-10 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9037873 

Testb1|a (refinement given correct calibration)            :  7.552425e-11 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6038159  (sd 0.090457) 

NULL Deviance                : 0.5585 (sd 0.0058812)

Fit Deviance                 : 0.31023 (sd 0.056142)

Explained Deviance           : 0.24827 (sd 0.056109)

Percent Deviance Explained   : 44.453 (sd 10.028)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.162 (sd 0.015492) 

Confusion Matrix: 

         observed

predicted    1    0

        1  162  201

        0   47 2196

AUC                          :  0.94447  (sd 0.020769) 

Percent Correctly Classified :  90.48029  (sd 1.504) 

Sensitivity                  :  0.7745238  (sd 0.11426) 

Specificity                  :  0.9160954  (sd 0.02072) 

Kappa                        :  0.5172162  (sd 0.047257) 

True Skill Statistic         :  0.6906192  (sd 0.10212) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.1014993  (sd 0.33549) 

Slope   (direction and variation in fit)                   :  0.9322373  (sd 0.31667) 

Testa0b1 (overall reliability of predictors)               :  0.376816  (sd 0.38515) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.5215378  (sd 0.23609) 

Testb1|a (refinement given correct calibration)            :  0.3672816  (sd 0.38356) 

Total time = 2.91 min

MARS

Random Forests

MARS Model Results

Data:

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

n(pres)                 =      209

n(abs)                  =      2397

n covariates considered =      6

total time for model fitting = 0.05min

Settings:

random seed used             : 123

mars degree                  : 1

mars penalty                 : 2

Summary of Model:

      nsubsets   gcv    rss

bio17       14 100.0  100.0

bio06       13  82.5   83.1

bio03       11  58.0   59.5

bio15       10  65.4>  66.2>

bio10        8  42.2   44.0

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.7128337 

NULL Deviance                : 0.55851

Fit Deviance                 : 0.23639

Explained Deviance           : 0.32212

Percent Deviance Explained   : 57.675

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.12 

Confusion Matrix: 

         observed

predicted    1    0

        1  188  230

        0   21 2167

AUC                          :  0.9675 

Percent Correctly Classified :  90.36838 

Sensitivity                  :  0.8995215 

Specificity                  :  0.9040467 

Kappa                        :  0.5517482 

True Skill Statistic         :  0.8035683 

   Calibration Statistics 

Intercept (general calibration)                            :  0.0002335318 

Slope   (direction and variation in fit)                   :  1.000633 

Testa0b1 (overall reliability of predictors)               :  0.9999504 

Testa0|b1(incorrect calibration given correct refinement)  :  0.997862 

Testb1|a (refinement given correct calibration)            :  0.9923434 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.7027542  (sd 0.070184) 

NULL Deviance                : 0.5585 (sd 0.0058812)

Fit Deviance                 : 0.27287 (sd 0.1122)

Explained Deviance           : 0.28563 (sd 0.11061)

Percent Deviance Explained   : 51.195 (sd 19.887)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.119 (sd 0.0056765) 

Confusion Matrix: 

         observed

predicted    1    0

        1  185  244

        0   24 2153

AUC                          :  0.96257  (sd 0.020855) 

Percent Correctly Classified :  89.71488  (sd 1.758) 

Sensitivity                  :  0.8854762  (sd 0.092911) 

Specificity                  :  0.898182  (sd 0.021217) 

Kappa                        :  0.5306853  (sd 0.053393) 

True Skill Statistic         :  0.7836581  (sd 0.085831) 

   Calibration Statistics 

Intercept (general calibration)                            :  0.08068983  (sd 0.53465) 

Slope   (direction and variation in fit)                   :  1.067486  (sd 0.35284) 

Testa0b1 (overall reliability of predictors)               :  0.3256605  (sd 0.31346) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.4759196  (sd 0.25681) 

Testb1|a (refinement given correct calibration)            :  0.297211  (sd 0.26501) 

Total time = 0.52 min

Random Forest Modeling Results

Data:

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

n(pres)                 =      209

n(abs)                  =      2397

n covariates considered =      6

total time for model fitting = 0.1min

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

bio10        32.8966 67.6009              64.1452          48.4177

bio06        40.6862 73.3698              63.5376          54.0887

bio15        30.8451 46.6611              56.4590          32.9069

bio03        29.3816 27.2944              51.1570          38.9075

bio17        29.1439 65.1121              49.0202          49.1576

di_all_model 20.8437  7.5533              22.6594          19.1783

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.7478534 

NULL Deviance                : 0.55851

Fit Deviance                 : 0.22932

Explained Deviance           : 0.32919

Percent Deviance Explained   : 58.941

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.1 

Confusion Matrix: 

         observed

predicted    1    0

        1  188  239

        0   21 2158

AUC                          :  0.9641 

Percent Correctly Classified :  90.02302 

Sensitivity                  :  0.8995215 

Specificity                  :  0.900292 

Kappa                        :  0.541858 

True Skill Statistic         :  0.7998136 

   Calibration Statistics 

Intercept (general calibration)                            :  0.04465583 

Slope   (direction and variation in fit)                   :  1.076547 

Testa0b1 (overall reliability of predictors)               :  0.4395513 

Testa0|b1(incorrect calibration given correct refinement)  :  0.7937312 

Testb1|a (refinement given correct calibration)            :  0.2093905 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.7414361  (sd 0.053919) 

NULL Deviance                : 0.5585 (sd 0.0058812)

Fit Deviance                 : 0.23519 (sd 0.04737)

Explained Deviance           : 0.32331 (sd 0.046022)

Percent Deviance Explained   : 57.908 (sd 8.3524)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.4725 (sd 0.036837) 

Confusion Matrix: 

         observed

predicted    1    0

        1  135   39

        0   74 2358

AUC                          :  0.96315  (sd 0.024734) 

Percent Correctly Classified :  95.66396  (sd 1.0392) 

Sensitivity                  :  0.6457143  (sd 0.10412) 

Specificity                  :  0.9837184  (sd 0.008917) 

Kappa                        :  0.6801662  (sd 0.077115) 

True Skill Statistic         :  0.6294327  (sd 0.10231) 

   Calibration Statistics 

Intercept (general calibration)                            :  0.1400177  (sd 0.43295) 

Slope   (direction and variation in fit)                   :  1.179537  (sd 0.36054) 

Testa0b1 (overall reliability of predictors)               :  0.3263459  (sd 0.24492) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.5463872  (sd 0.31873) 

Testb1|a (refinement given correct calibration)            :  0.2546577  (sd 0.23371) 

Total time = 1.31 min