search

nicholasjclark / MRFcov

Markov random fields with covariates
23 stars 5 forks source link

Error with indirect coefficient symmetrization 'new columns would leave holes after existing columns' #4

Closed nicholasjclark closed 6 years ago

nicholasjclark commented 6 years ago

I am trying to use the package to produce directly interpretable coefficients that allow me to determine the relative importance of species' interaction and environmental covariates in driving occurrence probabilities. I have been successful in exploring the regression coefficients and interpreting the results when I used my two species co-occurrence and three covariates. However, when I am trying to use 5 covariates, it gives me an error:

(Error in [<-.data.frame(*tmp*, , (n_nodes + (n_covariates * i) + (3 - : new columns would leave holes after existing columns.)

Is it a maximum of covariates number? I attach my data file in case you see any basic issue that I am not aware.

Thank you very much in advance. Congratulations for this fantastic package, I find really useful and interesting. I am looking forward to hearing from you.

nicholasjclark commented 6 years ago

This was an error sent to me by email from a user. This was caused by a bug in the previous version (1.0) that meant that indirect coefficients were not properly symmetrized. The new release (1.0.1) has amended this error