6.2 SDM结果展示

6.2.1 SDM响应曲线

##3.1 以网页形式查看结果:
rattler.me 
##3.2 查看刀切法构建的环境变量的敏感性(重要性)
plot(rattler.me)
##3.3  查看环境变量的因子的响应曲线;
response(rattler.me )

6.2.2 SDM概率分布图

概率连续分布图投影
##3.4 构筑分布概率图及绘制出图
rattler.pred <- predict(rattler.me, modelEnv)
plot(rattler.pred, main="Predicted Suitability")
map('worldHires', fill=FALSE, add=TRUE)
points(rattler$lon, rattler$lat, pch="+", cex=0.2)
##3.4.1 maxent参数中提供的案例:可以调参:
r <- predict(me, predictors, args=c("outputformat=raw"), progress='text', 
             filename='maxent_prediction.grd')
##3.4.2 maxent运行的结果投影在不同的数据上会产生不同的预测效果;
# 因此可以通过更改投影数据集实现数据转移;
ped1 <- predict(mod,clim_mask) # studyArea is the clipped rasters 
plot(ped1) # plot the continuous prediction
# 对比,换成全球栅格
ped2 <- predict(mod,clim)
plot(ped2)
# 投影在训练集中查看对应的概率值;
ped3 <- predict(mod,env_occ_train)
head(ped3)
概率分布图二值化Thresholds
thd1 <- threshold(mod_eval_train,stat="no_omission") # 0% omission rate 
thd2 <- threshold(mod_eval_train,stat="spec_sens") # highest TSS
thd3 <- threshold(mod_eval_train,stat="sensitivity",sensitivity=0.9) # 10% omission rate, i.e. sensitivity=0.9
thd4 <- threshold(mod_eval_train,stat="sensitivity",sensitivity=0.95) # 5% omission rate, i.e. sensitivity=0.95
# plotting points that are higher than the previously calculated thresholded value
plot(ped1>=thd1)

变量评估重要性

var.imp <- drop(get_variables_importance(PIPO.mod))
barplot(height = t(var.imp),
beside = TRUE,
horiz = TRUE,
xlab = "Variable Importance",
legend = c("GLM", "GAM", "ANN", "RF", "MAXENT"))

results matching ""

    No results matching ""