Page 43 - AI & Machine Learning for Beginners: A Guided Workbook

6. Evaluation Phase

• Purpose: Test the model on unseen data (the test set) to
predict its performance in real-world applications.
• Process:
o Evaluate the accuracy of predictions on the test
data.
o Ensure that the model, which has been trained on
the training set, generalizes well to new examples.

7. Parameter Tuning

• Hyperparameters: These include settings such as the
learning rate and the number of iterations.
o Learning Rate: Controls how drastically weights
and biases are adjusted at each training step.
o Iteration Count: Increasing the number of rounds
over the training data can improve accuracy, but
balance is crucial.
• Fine-Tuning: Adjust these hyperparameters based on
evaluation results. It’s common to experiment with
different values to achieve optimal performance.

This phase is crucial because even slight changes in these
parameters can have a substantial impact on overall model
accuracy and training time.

8. Deploying the Model

• Final Stage: After thorough training and evaluation, the
model is ready for deployment.
• Outcome: The model should correctly classify new
beverages as beer or wine, based on the learned
relationships between color and alcohol content.

This entire process—from defining the problem, collecting and
preparing data, selecting and training a model, to evaluating
and tuning it—illustrates the fundamental steps of training an
AI. It shows that effective training depends heavily on quality data,
41