Numerical Modeling Techniques GEOG 475 TAMU
Empirical curve fitting
Here is a curve that fits the trend, so lets just use the equation on the curve
Basically multiple regression
Linear regression
Variance in independent variable to predict the variance in the dependent variable
Can predict y by doing x
Can use 50-60 x-s to predict y
Has to be trained
Parametric stat two variables
Stronger the prob, stronger relationship, no relationship is no prob
Multiple regression
Parametric
Lots of variables
Has to be linear
Can transform to polynomial
Geographically-weighted regression
Prediction at local scale in multiple places to come up with what the __________ should be based on many locations
parameteric
Logistic regression
Exists or doesn't exist
Only two options
X or y; yes or no
1s and 0s and prob of 1s and 0s
Based on linear regression
Has to be trained
Cellular Automata
Basically AI
Rule-based
Suitability-site analysis (Criteria-based modeling)
Where is best for x (ranked)
How suitable is x for these uses/criteria
Empirical
Agent-based modeling
Model individuals or animals, how they interact, how to make decisions, and how the population responds
Numerical modeling