SCM

[#6662] BIOMOD_Modeling - ANN and MARS algorithms problem

Date:
2020-03-16 19:29
Priority:
3
State:
Open
Submitted by:
David Dolci (ddolci)
Assigned to:
Nobody (None)
BIOMOD Version:
biomod2
Product:
None
Operating System:
None
Version:
None
Severity:
None
Resolution:
None
Hardware:
None
URL:
https://www.youtube.com/watch?v=duBqW8l0e-U
Summary:
BIOMOD_Modeling - ANN and MARS algorithms problem

Detailed description
I’m having some problems with two algorithms when I launch the “BIOMOD_Modeling” function in BIOMOD2. ANN and MARS fail every single time and I don’t know why. The strange thing is that I was able to run both of them until two weeks ago. I’m wondering if the new version (3.4.6) has changed something related to these events.
I tried with different dataset: my data (composed by few occurrence points,), random data downloaded by GBIF (composed by hundreds of occurrence points) and also with the “GuloGulo” example. The result was always the same, a failure.
Could someone help me with this problem?!
David
(p.s: I tried to reinstall the old version of biomod2, but this hasn’t changed the results (i.e. ANN and MARS don’t work anymore also with old version. ;_;
p.p.s: fortunately, I saved the old “.RData” sessions in which ANN and MARS worked properly.)

################################_Reported errors_###################################
Video link to the GuloGulo attempt: https://youtu.be/duBqW8l0e-U
Model=Artificial Neural Network
5 Fold Cross Validation + 3 RepetitionsSetting levels: control = 0, case = 1
Setting direction: controls < cases
Setting levels: control = 0, case = 1
Setting direction: controls < cases
Setting levels: control = 0, case = 1
Setting direction: controls < cases
Setting levels: control = 0, case = 1
Setting direction: controls < cases
Setting levels: control = 0, case = 1
Setting direction: controls < cases
Setting levels: control = 0, case = 1
Setting direction: controls < cases
Setting levels: control = 0, case = 1
Setting direction: controls < cases
Setting levels: control = 0, case = 1
Setting direction: controls < cases
Setting levels: control = 0, case = 1
Model=Multiple Adaptive Regression Splines ( simple with no interaction )
Error in validObject(.Object) :
invalid class “MARS_biomod2_model” object: FALSE
In addition: Warning message:
glm.fit: fitted probabilities numerically 0 or 1 occurred
>


O.S: windows 10
r version 3.6.3
Rtools35.exe
biomod2 vers. 3.4.6

Comments:

No Comments Have Been Posted

Attached Files:

Changes

No Changes Have Been Made to This Item

Thanks to:
Vienna University of Economics and Business Powered By FusionForge