I have some evidence the ecol3 could be accepted into some models but seldom seems to be extremely informative. That being said, I am not certain (or at least question) how to interpret the predicted results and standard errors of estimate through broad scale review of the predictions when ecol3 is included.
Nevertheless, for this Issue, it is the fact that we have 30 with missing ecol3 (colored green below), which are the same locations as the missing bed permeability discussed in another Issue. We have 12 with ecol3_72 (colored blue below) but my model does not see that ecoregion at all and hence can not make predictions. Also, we have 30 with ecol3_37 (colored red below).
Now this code aside, I don't actually remove these watersheds for as I type this Issue, I am not using ecol3 in my GAMs, hence, this might not materially be an issue. However, they appear coastal and then way up at the distal tops of our doman---hence outright deletion for purposes of keeping our output tables close to having similar record lengths might be justification.
BLUE SYMBOLS = ecol3_72, RED SYMBOLS = ecol3_37, GREEN SYMBOLS: ecol3 == "nodata"
I have some evidence the ecol3 could be accepted into some models but seldom seems to be extremely informative. That being said, I am not certain (or at least question) how to interpret the predicted results and standard errors of estimate through broad scale review of the predictions when ecol3 is included.
Nevertheless, for this Issue, it is the fact that we have 30 with missing ecol3 (colored green below), which are the same locations as the missing bed permeability discussed in another Issue. We have 12 with ecol3_72 (colored blue below) but my model does not see that ecoregion at all and hence can not make predictions. Also, we have 30 with ecol3_37 (colored red below).
message("REMOVING ecol3_37, ecol3_72, nodata (Ecoregion)") length(spCOV$comid[spCOV$ecol3 == "ecol3_37"]) [1] 198 length(spCOV$comid[spCOV$ecol3 == "ecol3_72"]) [1] 12 length(spCOV$comid[spCOV$ecol3 == "nodata"]) [1] 30
Now this code aside, I don't actually remove these watersheds for as I type this Issue, I am not using ecol3 in my GAMs, hence, this might not materially be an issue. However, they appear coastal and then way up at the distal tops of our doman---hence outright deletion for purposes of keeping our output tables close to having similar record lengths might be justification.
BLUE SYMBOLS = ecol3_72, RED SYMBOLS = ecol3_37, GREEN SYMBOLS: ecol3 == "nodata"