Data Analysis & Reporting Flashcards
7 cards from real AML practice questions. Tap to flip, then mark Knew It or Still Learning โ missed cards come back until you master them.
Read the first 7 Data Analysis & Reporting flashcards as text
When analyzing a dataset with heavy right skew, which transformation is most commonly applied before modeling?
Answer: Log transformation
Log transformation compresses large values and reduces right skew, making distributions more symmetric for modeling.
In a confusion matrix for a binary classifier, which metric is computed as TP / (TP + FP)?
Answer: Precision
Precision measures the proportion of positive predictions that are actually positive: TP / (TP + FP).
A data analyst notices that two features have a Pearson correlation of 0.98. What is the primary concern?
Answer: Multicollinearity
A near-perfect correlation between two predictors indicates multicollinearity, which can destabilize regression coefficient estimates.
Which visualization is most appropriate for displaying the distribution of a continuous variable across multiple categories?
Answer: Violin plot
Violin plots combine a box plot with a kernel density estimate, effectively showing distribution shape across categories.
In time-series analysis, what does the autocorrelation function (ACF) measure?
Answer: Correlation of a series with its own lagged values
ACF measures the correlation between a time series and its own past values at various lag intervals.
When creating a dashboard KPI report, which principle ensures that the most critical metric is immediately visible?
Answer: Above-the-fold placement
Placing the most critical KPI above the fold ensures executives see it without scrolling, following dashboard design best practices.
A feature importance report shows that 'customer_id' ranks as the top predictor. What does this most likely indicate?
Answer: Data leakage from a unique identifier
A unique identifier ranking as top predictor typically signals data leakage, where the ID correlates with the target through improper data joining.