3 ms·
I'm surprised to see decision trees referenced: they are unsuitable for modeling real-world situations unless _every_ possible alternative can be determined afo
by giardini 18y ago
I'm surprised to see decision trees referenced: they are unsuitable for modeling real-world situations unless _every_ possible alternative can be determined aforehand, something simply impossible for terrorism (or for most human activity, for that matter).
- trapper 18y agoPossible scenarios can easily be abstracted. E.g. Building attack, car bomb, etc etc. They have the data.
- giardini 18y agoCertainly scenarios can be and are described but not all possible scenarios can be listed because they are infinite in number, And this is true in a non-trivial sense. When you say "They have the data." I assume you are speaking of statistics. But - These are rare events and statistical analysis is at it's limits, - The past is no predictor of the future. Example: the introduction and continued enhancement of IEDs in Iraq. Although he doesn't elaborate on my first point above, I recommend that you also read Nassim Nicholas Taleb's book, "The Black Swan".