Predictive or explanatory? Is that the question?

Background/Question/Methods

Remote sensing, satellites and GPS instruments are producing massive data sets that are used by ecologists and forest managers. Computer intensive techniques such as algorithmic and data mining procedures are often used to extract information from diverse large data sets these days. In this talk I will discuss whether ‘explanatory’ modeling techniques can also be used in the analysis of large geospatio-temporal data and whether explanatory models can give insights into the data not easily achieved by data mining or purely ‘predictive’ models.

Results/Conclusions

I will present examples on modeling and estimating effects of human and climate threats on forest health to illustrate these points. In conclusion "The hallmark of good science is that it uses models and "theories" but never believes them" (M. B. Wilk, in John Tukey, The future of data analysis, Ann. Math. Statist. 33, 1-67, p.7)