5
$\begingroup$

I have constructed two variable importance plots in r using random forests regression the dependent variable is wheat yield and the independent variables are topographic and vegetative parameters. I am wanting to know which topographic/vegetative variables have the most importance influence on wheat yield. I have done this for two years of data. The variable importance plot is shown below for both years. How do I interpret this? the node purity for the two years is different?

Also, how do I know which plot is most accurate? The variation explained in 2015 is 70%. while the variation explained in 2016 is 40% (the graph for 2016 looks more accurate though?) - please correct me if I am mistaken.

library(randomForest)
rf1 <- randomForest(Yield_2015~TWI +Upslope_length+ Curvature + Aspect+Slope+NDVI_2015+SAVI_2015,
                data=Topo_yield)
rf1$importance
varImpPlot(rf1,main="Important variables for Yield 2015")
rf2 <- randomForest(Yield_2016~TWI +Upslope_length+ Curvature + Aspect+Slope+NDVI_2016+SAVI_2016,
                data=Topo_yield_16)
varImpPlot(rf2,main="Important variables for Yield 2016")

Variable importance plot 2015 Variable importance plot 2016

Improve this question
$\endgroup$
0

0

Reset to default

Browse other questions tagged or ask your own question.