Final answer:
Evaluation on training data does not assess a model's ability to generalize to new data, which is essential to avoid over fitting. Techniques like cross-validation and regularization help mitigate over fitting, and thorough assessment beyond training data fitting is crucial for confidence in the model's predictions.
Step-by-step explanation:
Examining Overfitting: Evaluation on training data provides no assessment of how well the model generalizes to new, unseen data. This is a crucial aspect of model evaluation because a model that overfits may perform exceptionally well on the training data but fail to predict future observations accurately.
Overfitting occurs when the complexity of the model is such that it starts to capture the noise in the data rather than the underlying trend. To mitigate overfitting, various techniques such as cross-validation, regularization, and pruning of decision trees can be employed. It is also essential to evaluate the model on a separate test set that the model has not seen during the training phase.
This helps to understand the model's ability to generalize and hence its usefulness in making predictions about the real world. As a hypothesis, it is necessary for a model to be critically assessed on criteria such as its representation of the real world, limitations, and overall usefulness. While matching predictions with real-world observations is a positive sign, it does not conclusively prove the model's correctness.