The classic dataset for multiclass classification is the Iris dataset composed of three types of Iris flowers. This dataset is quite simple, very popular and using it to illustrate the performance of a platform as powerful as Amazon ML seems overkill. Luckily, there are another three class datasets composed of seeds. The seeds dataset is available at https://archive.ics.uci.edu/ml/datasets/seeds and of course on the GitHub repository accompanying this book (as well as the schema).
The seed dataset has 210 samples distributed evenly among three different seedTypes and seven attributes. The dataset has an ID, which must be set to categorical, all attributes are NUMERIC, and the target is the seedType. We ...
