Objective 2.1 Recognize Relative Impact of Data Quality and Size to Algorithms AIP-210 Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   A team has a large dataset with many mislabeled examples. Which statement best reflects the trade-off between data quality and quantity?

   • AThe mislabeled examples will likely introduce noise that can degrade model performance, so a smaller, high-quality dataset may be more effective.
   • BThe model will automatically correct mislabeled examples if the dataset is large enough.
   • CThe size of the dataset is the only factor that matters; quality has no impact on model performance.
   • DThe large dataset will always outperform a smaller, cleaner dataset because more data is always better.
   Reveal Solution Discussion Select an answer first
2. 7application · medium

   A team is training a model to predict customer lifetime value. They find that the 'income' feature has a small number of extreme values that are likely data entry errors (e.g., income of $1,000,000 for a student). They decide to remove these outliers. What is the most likely impact of this removal?

   • AThe model will overfit because the dataset becomes smaller
   • BThe model will have no change because outliers are automatically handled by the algorithm
   • CThe model will be less sensitive to extreme values and may generalize better to typical customers
   • DThe model will become biased because outliers contain important information
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   What is a potential downside of removing outliers from a training dataset?

   • AIt increases the size of the dataset, which can lead to overfitting.
   • BIt may remove genuine, informative data points that represent real-world variability.
   • CIt always increases the model's accuracy because outliers are always noise.
   • DIt has no effect on the model's performance.
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   What is the primary purpose of imputation in data preprocessing?

   • ATo normalize the data to a standard scale.
   • BTo increase the size of the dataset by generating synthetic samples.
   • CTo remove all rows that contain any missing values.
   • DTo replace missing values with estimated values so the dataset can be used for training.
   Reveal Solution Discussion Select an answer first
6. 10expert · hard

   A team is building a model to predict housing prices. They have a dataset with 10,000 records. The 'square_feet' feature has 5% missing values, and the 'price' feature has a few extreme outliers. They are considering two strategies: (1) drop all records with missing square_feet, then remove outliers from price; (2) impute missing square_feet with the median, then winsorize price at the 99th percentile. Which strategy is more likely to preserve model performance?

   • AStrategy 1, because imputation always introduces bias and should be avoided
   • BStrategy 1, because removing incomplete and outlier records ensures a clean dataset
   • CStrategy 2, because winsorizing is always better than removing outliers
   • DStrategy 2, because imputation retains data and winsorizing reduces outlier influence without losing records
   Reveal Solution Discussion Select an answer first