Data Issues DY0-001 Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A data scientist is analyzing sensor readings from a manufacturing process. A few readings are extremely high due to sensor malfunctions, but the rest are normal. These malfunction readings are not of interest. What is the best way to handle them?

   • AApply a log transformation to reduce their impact.
   • BReplace them with the mean of the normal readings.
   • CKeep them and treat them as valid data points.
   • DRemove them as outliers.
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A data scientist is modeling the relationship between a drug dosage and patient response. The response increases linearly up to a point, then levels off. A linear regression model shows a poor fit. What is the most appropriate next step?

   • AAdd a polynomial term, such as dosage squared, to the model.
   • BUse a simple linear regression with dosage only.
   • CRemove the plateau observations as outliers.
   • DConvert dosage to a binary variable (low vs. high).
   Reveal Solution Discussion Select an answer first
4. 18expert · hard

   A data scientist is building a fraud detection model. The dataset is highly imbalanced: 99.9% legitimate transactions and 0.1% fraudulent. The model currently achieves 99.9% accuracy but fails to catch most fraud. The business prioritizes catching fraud over minimizing false positives. What is the best approach?

   • AUse a larger dataset to include more fraud cases.
   • BUse SMOTE to oversample the fraud class and evaluate using recall or F1-score.
   • CUse accuracy as the evaluation metric and keep the current model.
   • DRemove the fraudulent transactions to balance the dataset.
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   Which of the following techniques is commonly used to address non-linearity in a dataset before applying a linear model?

   • ARemoving all outliers from the dataset
   • BIncreasing the number of training samples
   • CStandardizing the features to have zero mean and unit variance
   • DApplying a logarithmic transformation to the features
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   A data analyst is forecasting monthly ice cream sales. The data shows a clear seasonal pattern: sales peak in summer and are lowest in winter. However, the business also wants to understand the impact of a recent marketing campaign that ran for three months. The campaign data is only available at the weekly level. What is the most appropriate approach?

   • ADisaggregate the weekly campaign data to monthly using interpolation.
   • BUse monthly sales data and create a campaign dummy for the three months.
   • CKeep the monthly sales data and ignore the weekly campaign data.
   • DAggregate the sales data to weekly totals and include a campaign indicator variable.
   Reveal Solution Discussion Select an answer first