I recently presented at the Informs Roundtable Meeting on Data Mining. I will post several short individual items on useful learnings.
Data mining is about finding useful patterns in data, frequently from data that was not gathered explicitly to solve a current problem at hand. Modeling methods for data mining include neural nets, regression, genetic algorithms and classification approaches. Dr John Elder of Elder Research promotes the idea of trying a number of models and then combining the results in an ensemble result, the simplest approach would be to average the results ... a kind of Wisdom of Models, see his paper on why an ensemble of models can do better.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment