Define the Requirements of the Model

Regression modeling should start with defining the problem you are interested in solving. There are great challenges to modeling natural phenomena and taking problems from real situation may actually make it easier to have a viable model as managers need the best information they can get to make decisions each year for allocating resources. The phenomenon could be a measured variable, density of points, habitat suitability, and others. Examples include:

• Tree height
• Biomass of vegetation
• Distance to urban resources such as parks, hospitals, and stores
• Species habitat suitability (inc. ecological niche)
• Species distribution

The detailed requirements of a model include the uncertainty of the response variable including:

1. Spatial accuracy (bias) precision
2. Temporal accuracy and precision
3. Accuracy of the response values