Hands-On Exercise: Exploratory Testing and Exploratory Data Analysis (EDA) CT-AI Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   A tester is exploring an AI-based chatbot to find defects. Which action best exemplifies exploratory testing?

   • AEntering unusual or unexpected inputs and observing the chatbot's responses to guide further probing.
   • BExecuting a pre-written test script that checks 50 predefined user inputs.
   • CRunning a regression test suite to ensure no new defects were introduced.
   • DReviewing the chatbot's source code to find logic errors.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   A data scientist is performing EDA on a dataset for an image classification model. Which technique would be most appropriate to inspect the distribution of pixel intensity values?

   • ARunning a SQL query to join tables.
   • BWriting a unit test for the model's loss function.
   • CDeploying the model to a cloud server.
   • DCreating a histogram of pixel intensity values.
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A data scientist is performing EDA on a dataset for a house-price-prediction model. The dataset includes a 'location' feature with values like 'Downtown', 'Suburb', and 'Rural'. What is the most appropriate way to explore this feature during EDA?

   • ACalculate the mean and standard deviation of the location feature.
   • BRemove the location feature because it is categorical and cannot be used in a regression model.
   • CConvert the location feature to a numerical scale (e.g., 1, 2, 3) and plot a scatter plot against price.
   • DCreate a bar chart showing the frequency of each location category and examine the distribution of house prices within each category.
   Reveal Solution Discussion Select an answer first
5. 9expert · hard

   A team is developing a facial-recognition system for security access. Exploratory testing shows the system fails more often for individuals with darker skin tones. EDA reveals that the training dataset is predominantly composed of lighter-skinned individuals. The team has a strict deadline and cannot collect new data. What is the most appropriate course of action?

   • ASwitch to a different model architecture that is known to be less biased.
   • BReduce the model's sensitivity threshold to make it more likely to accept all individuals, reducing false rejections.
   • CUse data augmentation techniques (e.g., brightness/contrast adjustments) to create a more balanced training set, and use exploratory testing to validate the improvement on the affected demographic.
   • DDeploy the system as-is and document the known limitation in the release notes.
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A data scientist is performing EDA on a tabular dataset for a credit-risk model. The dataset contains a feature 'age' with some values recorded as 0 and some as 999. What is the most appropriate initial interpretation of these values during EDA?

   • ATreat 0 and 999 as likely placeholder or error values that need to be investigated before deciding how to handle them.
   • BAssume 0 represents newborns and 999 represents missing data, and proceed with modeling.
   • CImmediately remove all rows with age 0 or 999 to clean the dataset for modeling.
   • DReplace all 0 and 999 values with the median age of the dataset.
   Reveal Solution Discussion Select an answer first