Subject: Marine Geospatial Ecology Tools (MGET) help
Text archives
| From: | Jason Roberts <> |
|---|---|
| To: | Nere Zorrozua Ganboa <>, "" <> |
| Subject: | RE: [mget-help] Asking for help to use Marine Geospatial Ecology tools |
| Date: | Thu, 24 Jul 2014 02:31:08 +0000 |
| Accept-language: | en-US |
|
Dear Nere, Thanks for your interest in MGET. I can help you diagnose this problem, but I am away on holiday this week so I cannot look at it until next week. I am not sure I understand the problem completely, but when I have encountered similar problems in the past, there have been two usual causes: 1.
Predicting the model on data that were different than those used to fit the model. An example of this is to fit the model to data collected
in situ (e.g. measurements taken in the water when an animal was sighted), but then predict the data using rasters from remote sensing. Sometimes this will work but sometimes it can fail, due to differences in the two data sources. 2.
Fitting the model to one range of values but then predicting it on values outside that range. This is known as extrapolating. For example, the model
might be fitted to data that range from values 15-25, but then the model is predicted on values ranging from 10-30. Is the model reliable for the values in the range 10-15 or 25-30? Perhaps, or perhaps not. This can be risky. I hope that helps. If you send me some details on your model next week, such as the table you’re using to fit it and the complete output from the Fit GAM tool,
I may be able to offer better advice. Best regards, Jason From: Nere Zorrozua Ganboa [mailto:]
Dear Sir or Madam I am Nere Zorrozua, a student in the Universiy of the Basque Country (EHU-UPV) and I write to you to ask for help. I am doing the final master work about habitat suitability
modelling and I will be very grateful if you could help me. I am using Marine Geospatial Ecology Tools to make models, using 4 environmental variables' rasters to predict one biological point featured variable. It seems I do not have problems to fit a GAM
model using a formula and it seems the results of predicting GAM from table are correct too. But when I predict the model from rasters, the output raster has wrong values, being values of some million some of them. I am not able to find a solution to this
and I will be extremely thankful if you help me with this. Sorry becouse of my poor english and thank you very much for you help. Best regards, Nere Zorrozua |