EDA Methods DY0-001 Practice Questions (Page 3)

Questions 11–15

1. 11application · medium

   An analyst is working with a dataset where the target variable is 'loan_default' (binary). They have a categorical feature 'employment_type' with 12 categories and a numeric feature 'credit_score'. The analyst wants to determine which feature is more predictive of default. Which EDA comparison is most appropriate?

   • ACreate a histogram of credit_score and a pie chart of employment_type without referencing the target
   • BCalculate the Pearson correlation between employment_type (encoded) and credit_score
   • CCompare the default rate across employment_type categories using a bar chart, and compare the distribution of credit_score for defaulters vs. non-defaulters using box plots
   • DPlot a scatter plot of employment_type (encoded) vs. credit_score, colored by default status
   Reveal Solution Discussion Select an answer first
2. 12expert · hard

   A data analyst is exploring a dataset with two continuous variables, X and Y. A scatter plot shows a clear curved (U-shaped) relationship. The analyst computes the Pearson correlation coefficient and finds it is approximately 0. Which conclusion is most defensible?

   • AThe variables are completely unrelated, and further analysis is unnecessary
   • BThe variables have a strong negative relationship because the U-shape is symmetric
   • CThe Pearson correlation is invalid because the data must be normally distributed for it to be computed
   • DThe variables have no linear relationship, but a strong non-linear relationship may exist; Spearman correlation and a fitted curve should be examined
   Reveal Solution Discussion Select an answer first
4. 13expert · hard

   A data scientist is performing EDA on a dataset with a continuous target and several categorical features. They create a scatter plot of a numeric feature vs. the target and notice the points form distinct vertical bands. What is the most likely cause, and what is the appropriate next EDA step?

   • AThe scatter plot is overplotting due to too many points; a hexbin plot should be used instead
   • BThe numeric feature is actually discrete or has been rounded/grouped; a histogram or value-count table of the feature should be examined
   • CThe target variable has been log-transformed; the transformation should be reversed
   • DThere is a data leakage issue where the feature contains the target information
   Reveal Solution Discussion Select an answer first
5. 14expert · hard

   A data scientist is performing EDA on a dataset with a binary target and 100 features. They want to shortlist the most promising features for a predictive model. The dataset has some outliers and the relationships may be non-linear. Which feature-ranking approach during EDA is most robust for this shortlisting?

   • ACompute the Pearson correlation of each feature with the target
   • BTrain a full logistic regression model with all 100 features and rank by coefficient magnitude
   • CCompute the Area Under the ROC Curve (AUC) for each feature used individually to predict the target
   • DCount the number of missing values in each feature
   Reveal Solution Discussion Select an answer first
6. 15application · medium

   A marketing analyst has a dataset of 5,000 customers with a 'total_spend' variable that is heavily right-skewed, with a small number of high-spending outliers. The analyst needs to present the typical spending behavior to a non-technical stakeholder. Which chart and summary statistic combination is most appropriate?

   • AA box plot of total_spend with the median annotated
   • BA pie chart of total_spend by segment
   • CA bar chart of the mean total_spend by customer segment
   • DA histogram of total_spend with the mean annotated
   Reveal Solution Discussion Select an answer first