Species 058: Juniperus pinchotii

General Linear Models

Boosted Regression Trees

Generalized Linear Model Results

Data:

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

n(pres)                 =      178

n(abs)                  =      4822

n covariates considered =      5

total time for model fitting = 0.53min

Settings:

model family          : binomial

simplification method : AIC

Results:

number covariates in final model   : 18

Call:

glm(formula = response ~ bio01 + I(bio01^2) + I(di_all_model^2) + 

    di_all_model + I(bio08^2) + bio08 + I(bio02^2) + bio02 + 

    bio18 + I(bio18^2) + bio01:bio08 + bio01:bio02 + bio01:bio18 + 

    bio08:bio18 + bio01:di_all_model + di_all_model:bio02 + di_all_model:bio18 + 

    di_all_model:bio08, family = model.family, data = dat, weights = weight, 

    na.action = "na.exclude")

Deviance Residuals: 

    Min       1Q   Median       3Q      Max  

-1.7536  -0.0128   0.0000   0.0000   3.2831  

Coefficients:

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

(Intercept)        -1.630e+03  2.832e+02  -5.755 8.65e-09 ***

bio01               1.098e+01  1.500e+00   7.320 2.48e-13 ***

I(bio01^2)         -2.384e-02  2.939e-03  -8.110 5.04e-16 ***

I(di_all_model^2)  -8.574e-03  2.014e-03  -4.257 2.08e-05 ***

di_all_model       -5.278e+00  1.508e+00  -3.500 0.000466 ***

I(bio08^2)         -2.924e-03  1.201e-03  -2.434 0.014915 *  

bio08              -9.017e-01  6.458e-01  -1.396 0.162640    

I(bio02^2)         -2.069e-02  5.203e-03  -3.976 7.01e-05 ***

bio02               9.043e+00  2.174e+00   4.160 3.18e-05 ***

bio18               1.856e+00  4.221e-01   4.397 1.10e-05 ***

I(bio18^2)         -2.006e-03  4.283e-04  -4.684 2.81e-06 ***

bio01:bio08         8.686e-03  2.554e-03   3.401 0.000672 ***

bio01:bio02        -1.806e-02  4.004e-03  -4.511 6.46e-06 ***

bio01:bio18        -1.408e-02  2.013e-03  -6.995 2.65e-12 ***

bio08:bio18         4.970e-03  1.243e-03   3.998 6.38e-05 ***

bio01:di_all_model  1.774e-02  4.183e-03   4.240 2.23e-05 ***

di_all_model:bio02  1.744e-02  5.234e-03   3.332 0.000862 ***

di_all_model:bio18  4.463e-03  1.752e-03   2.547 0.010863 *  

di_all_model:bio08 -3.934e-03  2.077e-03  -1.894 0.058195 .  

---

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

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 1536.99  on 4999  degrees of freedom

Residual deviance:  595.84  on 4981  degrees of freedom

AIC: 633.84

Number of Fisher Scoring iterations: 13

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.6896366 

NULL Deviance                : 0.3074

Fit Deviance                 : 0.11917

Explained Deviance           : 0.18823

Percent Deviance Explained   : 61.233

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.06 

Confusion Matrix: 

         observed

predicted    1    0

        1  169  323

        0    9 4499

AUC                          :  0.9816 

Percent Correctly Classified :  93.36 

Sensitivity                  :  0.9494382 

Specificity                  :  0.9330153 

Kappa                        :  0.4771403 

True Skill Statistic         :  0.8824535 

   Calibration Statistics 

Intercept (general calibration)                            :  0.001258352 

Slope   (direction and variation in fit)                   :  1.002319 

Testa0b1 (overall reliability of predictors)               :  0.9994951 

Testa0|b1(incorrect calibration given correct refinement)  :  0.995886 

Testb1|a (refinement given correct calibration)            :  0.9749818 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6610139  (sd 0.096841) 

NULL Deviance                : 0.30738 (sd 0.0053556)

Fit Deviance                 : 0.13031 (sd 0.036223)

Explained Deviance           : 0.17707 (sd 0.033972)

Percent Deviance Explained   : 57.687 (sd 11.432)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.062 (sd 0.0063246) 

Confusion Matrix: 

         observed

predicted    1    0

        1  162  327

        0   16 4495

AUC                          :  0.97739  (sd 0.013843) 

Percent Correctly Classified :  93.14023  (sd 0.82044) 

Sensitivity                  :  0.9111111  (sd 0.098688) 

Specificity                  :  0.9321856  (sd 0.0085865) 

Kappa                        :  0.4580485  (sd 0.044616) 

True Skill Statistic         :  0.8432967  (sd 0.097069) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.0385575  (sd 0.26918) 

Slope   (direction and variation in fit)                   :  1.046194  (sd 0.4022) 

Testa0b1 (overall reliability of predictors)               :  0.408  (sd 0.3429) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.6497638  (sd 0.25801) 

Testb1|a (refinement given correct calibration)            :  0.3179826  (sd 0.34077) 

Total time = 5.88 min

Boosted Regression Tree Modeling Results

Data:

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

n(pres)                 =      178

n(abs)                  =      4822

n covariates considered =      5

total time for model fitting = 0.2min

Settings:

random seed used             : 19680828

tree complexity              : 9

learning rate                : 0.087

n(trees)                     : 300

model simplification         : cross-validation

n folds                      : 3

n covariates in final model  : 4

Relative influence of predictors in final model:

          Var  rel.inf

        bio02 33.76256

        bio18 24.11633

        bio01 23.87844

 di_all_model 18.24268

Important interactions in final model:

 v1 name1 v2 name2

  3 bio18  2 bio02

  3 bio18  1 bio01

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.96668 

NULL Deviance                : 0.3074

Fit Deviance                 : 0.024769

Explained Deviance           : 0.28263

Percent Deviance Explained   : 91.942

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.18 

Confusion Matrix: 

         observed

predicted    1    0

        1  177   27

        0    1 4795

AUC                          :  0.9998 

Percent Correctly Classified :  99.44 

Sensitivity                  :  0.994382 

Specificity                  :  0.9944007 

Kappa                        :  0.9238044 

True Skill Statistic         :  0.9887827 

   Calibration Statistics 

Intercept (general calibration)                            :  1.078933 

Slope   (direction and variation in fit)                   :  2.918262 

Testa0b1 (overall reliability of predictors)               :  3.330669e-16 

Testa0|b1(incorrect calibration given correct refinement)  :  0.8572926 

Testb1|a (refinement given correct calibration)            :  0 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.6589119  (sd 0.14623) 

NULL Deviance                : 0.30738 (sd 0.0053556)

Fit Deviance                 : 0.14358 (sd 0.056909)

Explained Deviance           : 0.16379 (sd 0.05411)

Percent Deviance Explained   : 53.44 (sd 18.103)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.2085 (sd 0.041099) 

Confusion Matrix: 

         observed

predicted    1    0

        1  122  113

        0   56 4709

AUC                          :  0.97245  (sd 0.023165) 

Percent Correctly Classified :  96.62072  (sd 2.0721) 

Sensitivity                  :  0.6866013  (sd 0.10606) 

Specificity                  :  0.976565  (sd 0.019315) 

Kappa                        :  0.5958024  (sd 0.1439) 

True Skill Statistic         :  0.6631663  (sd 0.11622) 

   Calibration Statistics 

Intercept (general calibration)                            :  -0.192374  (sd 0.54173) 

Slope   (direction and variation in fit)                   :  0.7749921  (sd 0.17203) 

Testa0b1 (overall reliability of predictors)               :  0.2353988  (sd 0.31764) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.4572494  (sd 0.3143) 

Testb1|a (refinement given correct calibration)            :  0.2546187  (sd 0.34589) 

Total time = 2.72 min

MARS

Random Forests

MARS Model Results

Data:

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

n(pres)                 =      178

n(abs)                  =      4822

n covariates considered =      5

total time for model fitting = 0.04min

Settings:

random seed used             : 123

mars degree                  : 1

mars penalty                 : 2

Summary of Model:

      nsubsets   gcv    rss

bio18       11 100.0  100.0

bio02       10  93.2   93.2

bio08        7  52.8   54.0

bio01        6  43.0   44.5

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.5707064 

NULL Deviance                : 0.3074

Fit Deviance                 : 0.16284

Explained Deviance           : 0.14456

Percent Deviance Explained   : 47.026

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.04 

Confusion Matrix: 

         observed

predicted    1    0

        1  157  654

        0   21 4168

AUC                          :  0.9552 

Percent Correctly Classified :  86.5 

Sensitivity                  :  0.8820225 

Specificity                  :  0.8643716 

Kappa                        :  0.2751731 

True Skill Statistic         :  0.7463941 

   Calibration Statistics 

Intercept (general calibration)                            :  0.001257747 

Slope   (direction and variation in fit)                   :  1.00116 

Testa0b1 (overall reliability of predictors)               :  0.9998125 

Testa0|b1(incorrect calibration given correct refinement)  :  0.9963392 

Testb1|a (refinement given correct calibration)            :  0.9849871 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.5586426  (sd 0.14748) 

NULL Deviance                : 0.30738 (sd 0.0053556)

Fit Deviance                 : 0.16807 (sd 0.047253)

Explained Deviance           : 0.13931 (sd 0.045421)

Percent Deviance Explained   : 45.403 (sd 15.019)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.044 (sd 0.0069921) 

Confusion Matrix: 

         observed

predicted    1    0

        1  158  571

        0   20 4251

AUC                          :  0.95295  (sd 0.029494) 

Percent Correctly Classified :  88.18182  (sd 2.3881) 

Sensitivity                  :  0.8879085  (sd 0.082603) 

Specificity                  :  0.8815958  (sd 0.023402) 

Kappa                        :  0.3149052  (sd 0.065632) 

True Skill Statistic         :  0.7695043  (sd 0.094323) 

   Calibration Statistics 

Intercept (general calibration)                            :  0.01241452  (sd 0.56978) 

Slope   (direction and variation in fit)                   :  1.107221  (sd 0.52581) 

Testa0b1 (overall reliability of predictors)               :  0.4171244  (sd 0.39821) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.6758337  (sd 0.24734) 

Testb1|a (refinement given correct calibration)            :  0.3804735  (sd 0.37723) 

Total time = 0.47 min

Random Forest Modeling Results

Data:

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

n(pres)                 =      178

n(abs)                  =      4822

n covariates considered =      5

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

di_all_model 44.7719 64.0433              71.1545          35.0599

bio01        45.3397 73.0138              62.3282          39.3629

bio18        45.6033 99.7031              58.6356          50.1085

bio02        45.2595 72.0722              57.5019          49.3048

bio08        47.4259 43.0892              55.9959          41.8027

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

Evaluation Statistics applied to train split:

Correlation Coefficient      : 0.7434644 

NULL Deviance                : 0.3074

Fit Deviance                 : 0.1102

Explained Deviance           : 0.1972

Percent Deviance Explained   : 64.152

  Threshold Methods based on Sens=Spec 

Threshold                    :  0.05 

Confusion Matrix: 

         observed

predicted    1    0

        1  166  307

        0   12 4515

AUC                          :  0.983 

Percent Correctly Classified :  93.62 

Sensitivity                  :  0.9325843 

Specificity                  :  0.9363335 

Kappa                        :  0.4832522 

True Skill Statistic         :  0.8689177 

   Calibration Statistics 

Intercept (general calibration)                            :  0.1305058 

Slope   (direction and variation in fit)                   :  1.121859 

Testa0b1 (overall reliability of predictors)               :  0.08414175 

Testa0|b1(incorrect calibration given correct refinement)  :  0.6135587 

Testb1|a (refinement given correct calibration)            :  0.03024206 

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

Evaluation Statistics applied to crossValidation split:

Correlation Coefficient      : 0.7308237  (sd 0.091508) 

NULL Deviance                : 0.30738 (sd 0.0053556)

Fit Deviance                 : 0.1131 (sd 0.030609)

Explained Deviance           : 0.19427 (sd 0.02806)

Percent Deviance Explained   : 63.28 (sd 9.644)

  Threshold Methods based on Sens=Spec 

Mean Threshold               : 0.482 (sd 0.012517) 

Confusion Matrix: 

         observed

predicted    1    0

        1  101   33

        0   77 4789

AUC                          :  0.97724  (sd 0.016134) 

Percent Correctly Classified :  97.80032  (sd 0.6921) 

Sensitivity                  :  0.5689542  (sd 0.13772) 

Specificity                  :  0.9931565  (sd 0.0050855) 

Kappa                        :  0.6335445  (sd 0.12092) 

True Skill Statistic         :  0.5621108  (sd 0.13752) 

   Calibration Statistics 

Intercept (general calibration)                            :  0.1886707  (sd 0.57169) 

Slope   (direction and variation in fit)                   :  1.156259  (sd 0.42104) 

Testa0b1 (overall reliability of predictors)               :  0.4838448  (sd 0.30126) 

Testa0|b1(incorrect calibration given correct refinement)  :  0.6064732  (sd 0.20757) 

Testb1|a (refinement given correct calibration)            :  0.3963197  (sd 0.33748) 

Total time = 1.56 min