Hands-On Exercise: Exploratory Testing and Exploratory Data Analysis (EDA) CT-AI Practice Questions (Page 5)

Questions 21–25

1. 21application · medium

   A team is building a computer-vision model to detect manufacturing defects on a production line. The dataset contains 50,000 images, but the team suspects the images may have been collected under inconsistent lighting conditions. During exploratory data analysis, which action would most directly reveal whether lighting variability is likely to cause systematic bias in the model?

   • ACompute the overall accuracy of a pre-trained model on the full dataset to establish a baseline performance metric.
   • BRemove all images with extreme pixel values to ensure the dataset contains only 'normal' lighting conditions.
   • CPlot the distribution of pixel-intensity statistics (e.g., mean and standard deviation) across images and compare them across different production shifts.
   • DCount the number of images in each defect class to verify the dataset is balanced before training.
   Reveal Solution Discussion Select an answer first
2. 22application · medium

   During exploratory testing of an image-recognition system, a tester finds that the system misclassifies images of dogs when they are partially obscured by a shadow. What is the most valuable information to record in the defect report?

   • AA suggestion to retrain the model with more dog images.
   • BThe model's architecture and training hyperparameters.
   • CA general statement that the system is unreliable for animal images.
   • DThe specific image, the expected output, the actual output, and the environmental conditions (e.g., shadow) that triggered the failure.
   Reveal Solution Discussion Select an answer first
4. 23foundation · easy

   How can exploratory testing and EDA be integrated to jointly identify issues in an AI system?

   • ABy performing EDA first and then never using exploratory testing.
   • BBy running EDA only on the test data and exploratory testing only on the training data.
   • CBy using exploratory testing to generate synthetic data for EDA.
   • DBy using EDA to find data anomalies and exploratory testing to find system behaviors that may be caused by those anomalies.
   Reveal Solution Discussion Select an answer first
5. 24foundation · easy

   During EDA on a tabular dataset for a regression model, a data scientist notices that the 'age' column contains several negative values. What should be the immediate next step in EDA?

   • AIgnore the negative values because they are rare.
   • BInvestigate the source of the negative values to determine if they are data entry errors or legitimate values.
   • CImmediately delete all rows with negative age values.
   • DReplace all negative values with the mean age.
   Reveal Solution Discussion Select an answer first
6. 25foundation · easy

   Which of the following is a key benefit of applying exploratory testing to an AI-based system?

   • AIt guarantees that the AI model is 100% accurate.
   • BIt provides a complete, reproducible set of test cases for every scenario.
   • CIt eliminates the need for any other testing techniques.
   • DIt allows testers to discover defects that are difficult to anticipate from specifications.
   Reveal Solution Discussion Select an answer first