[Machine Learning is the] field of study that gives computers the ability to learn without being explicitly programmed.

—Arthur Samuel, 1959

General Approach

Step 1: Input Data + Output Data (Training Stage)

• Input data: The information you give to the model (e.g., height, weight, age).
• Output data: The answer you want the model to learn to predict (e.g., salary).
• A model is like a formula or algorithm that tries to find relationships between inputs and outputs.
• Model with hyperparameters:
  • Hyperparameters are settings you choose before training (i.e. learning rate, model complexity). This is not to be confused with the model’s internal parameters.
  • I will explain more about what these hyperparameters do in the later topics.

Step 2: ML Framework (Training Process)

• The ML framework (i.e. Scikit-learn, TensorFlow, or PyTorch) uses the input-output pairs to adjust the model's internal parameters.
• Training process: this is where the model "gets strong" by learning patterns in data.
• In the context of regression, the internal parameters are the coefficients (also called weights) and the bias/intercept.

Step 3: Trained Model (Prediction Stage)

• Now the model has trained parameters (weights/coefficients) and has learned how to map input → output.
• You can give it new input, and it will predict the output (e.g., predict someone's salary based on unseen data). This input can be the test data.
• This trained model is now ready to generalize to real-world cases.
• Basically, we always follow the flow of train → validation → test → deployment