Summary

The chapter began with an introduction to some of the most important datasets that will be used in the rest of the book. The datasets covered a range of analytical problems including classification, regression, time series, survival, clustering, and a dataset in which identifying an outlier is important. Important families of classification models were then introduced in the statistical/machine learning models section. Following the introduction of a variety of models, we immediately saw the shortcoming, in that we don't have a model for all seasons. Model performance varies from dataset to dataset. Depending on the initialization, the performance of certain models (such as neural networks) is affected. Consequently, there is a need to find a way to ensure that the models can be improved upon in most scenarios.

This paves the way for the ensemble method, which forms the title of this book. We will elaborate on this method in the rest of the book. This chapter closed with quick statistical tests that will help in carrying out model comparisons. Resampling forms the core of ensemble methods, and we will look at the important jackknife and bootstrap methods in the next chapter.