AWS Machine Learning

If you dont have business problem, then you can’t possibiliy come up with machine learning algorithm and model to answer what you are trying figure out.


What is a model?

Model is a trained algorithm which is

  • Used to identify patterns in your data, and
  • Does not require explicit, manually set rules.


Learn by identifying data that is already been labeled. We have/had data to train


  • Binary (Yes or No)
  • Multiclass Classfication (like tags. Ex: possible answers for call center - “IT Support”, “Returns”, “Accounting”)
  • Regression problem (Range)


There is no existing data points to train. We just have raw data. The machine has to uncover and create the labels itself.

  • use data grouping.
  • detects annomolies.

Reinforcement Learning

Learning through trial and error.