I would encourage you to try more than datasets (actually many many datasets) before drawing conclusions on accuracy. Here is our analysis of Google Cloud AutoML versus Auger.AI, Azure AutoML, H20, TPOT. https://medium.com/augerai/auger-outperforms-competitors-on-their-own-chosen-datasets-the-google-update-cd83b6abcc2e

The decision to include SageMaker is curious as its just hyperparameter optimization.