Model Evaluation

Model evaluation is the process of evaluating the performance of a Machine Learning model on a test set, as it changes in environment, use cases, and adaptations to changes in input data .

Types:

• Performance test: using test data
• Robustness test: using random or false data
• Model explainability
• Result evaluation

Notes:

• models should not be treated as finished projects, but as cycles.
• Model Compression Techniques can be used as a part of model evaluation before deploying the model.
• While Model Validation and Model Evaluation are both essential in creating any analytical Data Pipeline, they should not be mixed.