Hands-On Exercise: Evaluate the Created ML Model CT-AI Practice Questions (Page 4)

Questions 16–20

1. 16foundation · easy

   In a multi-class classification problem, which metric is commonly used to evaluate overall model performance by averaging class-wise F1-scores without considering class imbalance?

   • AMacro-averaged F1-score
   • BAccuracy
   • CWeighted F1-score
   • DMicro-averaged F1-score
   Reveal Solution Discussion Select an answer first
2. 17foundation · easy

   What is the primary purpose of documenting the evaluation process and findings for an ML model?

   • ATo increase the model's accuracy by recording the evaluation steps.
   • BTo replace the need for testing the model in production.
   • CTo satisfy legal requirements that all ML models must be documented.
   • DTo provide a record that supports reproducibility, transparency, and informed decision-making.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A team evaluated a multiclass model for classifying news articles into topics. The confusion matrix shows that the model often confuses 'sports' with 'entertainment'. Which metric would best quantify the model's ability to correctly identify 'sports' articles?

   • AThe F1-score for the 'entertainment' class
   • BPrecision for the 'entertainment' class
   • CRecall for the 'sports' class
   • DOverall accuracy
   Reveal Solution Discussion Select an answer first
5. 19application · medium

   After evaluating a sentiment analysis model, a team must document the findings for stakeholders who are not familiar with machine learning. Which approach best communicates the model's effectiveness?

   • AOnly report the overall accuracy, since it is the simplest metric.
   • BProvide a summary of key metrics with plain-language explanations and a few illustrative examples.
   • CShare the raw evaluation logs and let stakeholders draw their own conclusions.
   • DInclude the full confusion matrix and all metric formulas in the report.
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A team evaluated a multi-class model for classifying customer support tickets into categories. The confusion matrix shows that the model often confuses 'billing' with 'account' issues. Which metric would best quantify the model's ability to correctly identify 'billing' tickets?

   • AThe F1-score for the 'account' class
   • BRecall for the 'billing' class
   • CPrecision for the 'account' class
   • DOverall accuracy
   Reveal Solution Discussion Select an answer first