Amazon Machine Learning
Developer Guide (Version Latest)

Improving Model Accuracy

Obtaining a ML model that matches your needs usually involves iterating through this ML process and trying out a few variations. You might not get a very predictive model in the first iteration, or you might want to improve your model to get even better predictions. To improve performance, you could iterate through these steps:

  1. Collect data: Increase the number of training examples

  2. Feature processing: Add more variables and better feature processing

  3. Model parameter tuning: Consider alternate values for the training parameters used by your learning algorithm