MayaGueguen
commented
2 months ago Hello Shree,
Here are some insights that might help you understand what's going on and interpret your response' curves :sun_with_face:
For response curves, you want to check how the predicted value evolves in function of one variable. So when moving along the range of your variable of interest, the other variables are fixed to an average or median value so the variation you see in predicted value is only due to the variation in your variable of interest.
:warning: Keep in mind that this is not a perfect way to watch how your predictions evolve, as the more variables you have, the more the signal might be smoothed / hidden.
Let's take your bio12 variable for example. What the curve means is that :
- WHEN all other variables (so 10 here) are fixed to their MEAN value, predictions show their maximum value (here up to 0.2) when
bio12is ranging between 1700 and 2000. - BUT it is highly possible that you don't have any site actually matching these theoretical conditions (all 10 variables to their mean and precip at 1900)
On the contrary, if you look at elevation, it is more determinant (but less precise) as when all other 10 variables are fixed to their mean, predictions are higher as soon as elevation is above 800m.
:information_source: You might want to combine this information with variables' importance, to have another insight about which variables impact the most your predictions.
As for your second graphic, you apply a smooth function to all response' curves obtained for bio1 variable across all single models :
- the scale between 0 and 0.1 is understandable by the fact that it seems that a lot of the single response' curves are straight lines near to 0, hence dragging down the smoothed curve
- you might want to select only single models having good evaluation values
- as mentioned above, you might want to cross-check with variables' importance
Hope it helps, Maya
Hello again. I am trying to plot response curve of contribution of predictor variable in ensemble model. As we can see in the figure, shouldn't the value range between 0-1. Atleast thats the case when generating response curve for individual models.
.
Also i tried using this code for generating response curve
b <- bm_PlotResponseCurves( myBiomodModelOut, models.chosen = get_built_models(myBiomodModelOut, run = 'RUN1'), new.env = get_formal_data(myBiomodModelOut, "expl.var"), show.variables = c('bio1'), fixed.var = "mean", do.bivariate = FALSE, do.plot = TRUE, do.progress = TRUE )
abc <- b$tab
abc %>% group_by(expl.name) %>% ggplot(aes(expl.val, pred.val))+ geom_smooth() + facet_wrap(~ expl.name, scales = "free_x")
and i have same problem like the pred.val ranges between 0 - 0.1 whereas when the result from bm_plotresponsecurve has value upto 1.