Challenges in Data Preparation CT-AI Practice Questions (Page 6)

Questions 26–28

1. 26application · medium

   A data scientist is working on a project to predict whether a customer will respond to a marketing campaign. The dataset includes a 'date' column with the date of the last purchase. The scientist wants to create a feature that captures the recency of the purchase. What is the most appropriate feature to create?

   • AThe day of the week of the last purchase.
   • BThe number of days since the last purchase relative to the current date.
   • CThe month of the last purchase.
   • DThe year of the last purchase.
   Reveal Solution Discussion Select an answer first
2. 27application · medium

   A company is building a model to approve loan applications. The training data is collected from past loan decisions, which were made by human loan officers. The model is found to deny loans to a higher proportion of applicants from a certain ethnic group. What is the most likely source of this bias?

   • AThe training data contains historical biases, as past loan decisions may have been discriminatory.
   • BThe model is overfitting to the training data because of too many features.
   • CThe model is using a feature that is a proxy for ethnicity, such as zip code.
   • DThe model is not using enough training data, so it cannot learn the correct patterns.
   Reveal Solution Discussion Select an answer first
3. 28foundation · easy

   Which of the following is an example of bias in data preparation that could lead to unfair outcomes?

   • ARemoving duplicate records from the dataset
   • BScaling features to a similar range
   • CSplitting data randomly into train and test sets
   • DUsing a dataset where one demographic group is overrepresented in the positive class
   Reveal Solution Discussion Select an answer first