A quantitative assessment of site level factors - Chukar

## Loading required package: sp

## Checking rgeos availability: TRUE

## OGR data source with driver: ESRI Shapefile 
## Source: "/Users/austinsmith/Documents/SDM_spatial_data/Bird_life_galliformes_fgip/Alectoris_chukar/Alectoris_chukar.shp", layer: "Alectoris_chukar"
## with 3 features
## It has 18 fields
## Integer64 fields read as strings:  OBJECTID SISID PRESENCE ORIGIN SEASONAL

## OGR data source with driver: ESRI Shapefile 
## Source: "/Users/austinsmith/Downloads/ne_110m_admin_1_states_provinces/ne_110m_admin_1_states_provinces.shp", layer: "ne_110m_admin_1_states_provinces"
## with 51 features
## It has 83 fields
## Integer64 fields read as strings:  ne_id

## Loading required namespace: rgeos

AUC scores from cross-validaion of models

##        Fold_1 Fold_2 Fold_3 Fold_4 Fold_5
## ANN     0.997  0.997  0.997  0.996  0.996
## GBM     0.988  0.991  0.995  0.991  0.987
## MaxEnt  0.992  0.999  0.999  0.997  0.998
## RF      0.997  0.998  0.998  0.999  0.998
## SVM     0.914  0.996  0.996  0.994  0.996

## Loading required package: ggplot2

## Loading required package: magrittr

## 
## Attaching package: 'magrittr'

## The following object is masked from 'package:raster':
## 
##     extract

## 
## Attaching package: 'ggpubr'

## The following object is masked from 'package:raster':
## 
##     rotate

Algorithms

There are three types of ensemble models below. * Mean-value - calcuated, unweighted mean value for each cell in the raster (traditional) * Voting-polls(VP) - allows us to see how many folds voted on the raster cell * Unanimous decision - shows which location all folds agreed on

Artificial Neural Network

Insert discription of ann.

Predictions - regression

Fold 1

Fold 2

Fold 3

Fold 4

Fold 5

Ensemble - Mean

Predictions - dichotomy

Fold 1

Suitability threshhold = 0.0347402

Fold 2

Suitability threshhold = 0.0641611

Fold 3

Suitability threshhold = 0.0951801

Fold 4

Suitability threshhold = 0.0495876

Fold 5

Suitability threshhold = 0.04054

Ensemble - Mean

Suitability threshhold = 0.0568418

Ensemble - VP

Ensemble - UD

Boosted Trees

Insert discription of boosted trees.

Predictions - regression

Fold 1

Fold 2

Fold 3

Fold 4

Fold 5

Ensemble - Mean

Predictions - dichotomy

Fold 1

Suitability threshhold = -4.2633938

Fold 2

Suitability threshhold = -4.1765705

Fold 3

Suitability threshhold = -3.3434323

Fold 4

Suitability threshhold = -3.5482324

Fold 5

Suitability threshhold = -4.2504879

Ensemble - Mean

Suitability threshhold = -3.9164234

Ensemble - VP

Ensemble - UD

Maximum Entropy

Insert discription of maxent.

Predictions - regression

Fold 1

Fold 2

Fold 3

Fold 4

Fold 5

Ensemble - Mean

Predictions - dichotomy

Fold 1

Suitability threshhold = 0.218207

Fold 2

Suitability threshhold = 0.1402118

Fold 3

Suitability threshhold = 0.3381145

Fold 4

Suitability threshhold = 0.1019353

Fold 5

Suitability threshhold = 0.2440821

Ensemble - Mean

Suitability threshhold = -3.9164234

Ensemble - VP

Ensemble - UD

Random Forest

Insert discription of random forest.

Predictions - regression

Fold 1

Fold 2

Fold 3

Fold 4

Fold 5

Ensemble - Mean

Predictions - dichotomy

Fold 1

Suitability threshhold = 0.0303667

Fold 2

Suitability threshhold = 0.0662333

Fold 3

Suitability threshhold = 0.1232

Fold 4

Suitability threshhold = 0.2642475

Fold 5

Suitability threshhold = 0.1179

Ensemble -Mean

Suitability threshhold = 0.1203895

Ensemble - VP

Ensemble - UD

Support Vector Machine

Insert discription of support vector machines

Predictions - regression

Fold 1

Fold 2

Fold 3

Fold 4

Fold 5

Ensemble - Mean

Predictions - dichotomy

Fold 1

Suitability threshhold = 0.0085674

Fold 2

Suitability threshhold = 0.0087646

Fold 3

Suitability threshhold = 0.014438

Fold 4

Suitability threshhold = 0.0108407

Fold 5

Suitability threshhold = 0.0101643

Ensemble - Mean

Suitability threshhold = 0.010555

Ensemble - VP

Ensemble - UD

Collective Models

Final Model: Ensemble

Predictions - regression

Predictions - dichotomy

Predictions - VP

Predictions - UD

plot(crop( ensemble_sum , states) == 5, xlim = c( left, right ), ylim =c( bottom , top ))
plot(states, add = T)
plot(naturalized, add = T, border = "red")
legend("bottomright", c( "True range"),col=c( "red"), lty =1 )

All fold democracy

Predictions - VP

all_folds <- sum( ann_sum, 
                gbm_sum,
                maxent_sum,
                rf_sum, 
                svm_sum)

plot(crop( all_folds , states), xlim = c( left, right ), ylim =c( bottom , top ))
plot(states, add = T)
plot(naturalized, add = T, border = "red")
legend("bottomright", c( "True range"),col=c( "red"), lty =1 )

Predictions - UD

plot(crop( all_folds , states) == 25, xlim = c( left, right ), ylim =c( bottom , top ))
plot(states, add = T)
plot(naturalized, add = T, border = "red")
legend("bottomright", c( "True range"),col=c( "red"), lty =1 )