Ensemble modelling or selecting the best model: Many could be better than one

摘要: In the course of data modelling, many models could be created. Much work has been done on formulating guidelines for model selection. However, by and large, these are conservative or too specific. Instead using general guidelines, selected a particular task based statistical tests. When selecting one model, others discarded. losing potential sources information, combined to yield better performance. We review basics selection combination discuss their differences. Two examples opportunistic principled combinations presented. The first demonstrates that mediocre quality significantly latter is main contribution paper; it describes illustrates novel heuristic approach called SG(k-NN) ensemble generation good-quality diverse can even improve excellent models.