Visualization of Multivariate Data DATA-SCIENCE-OPTIMIZE Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A data analyst at a manufacturing plant has a dataset with 12 numeric process parameters (temperature, pressure, speed, etc.) recorded for each production batch. The analyst wants to quickly identify which pairs of parameters show the strongest linear relationships to prioritize root-cause analysis of quality defects. The dataset has no missing values and fits comfortably in memory. Which visualization approach is most appropriate as the first step?

   • ADisplay a heatmap of the raw parameter values.
   • BCreate a single scatter plot of temperature against pressure.
   • CGenerate a scatterplot matrix of all 12 parameters.
   • DBuild a parallel coordinates plot with all 12 parameters.
   Reveal Solution Discussion Select an answer first
2. 17expert · hard

   A data analyst has a dataset with 8 numeric variables and 5,000 observations. The analyst wants to identify clusters of observations and understand which variables drive the separation between clusters. The analyst is considering a scatterplot matrix and a parallel coordinates plot. Which consideration is most important when choosing between them?

   • AA scatterplot matrix shows pairwise relationships, while parallel coordinates show the full multi-variable profile of each observation.
   • BBoth plots are equivalent for this task.
   • CA scatterplot matrix is always better for cluster identification.
   • DParallel coordinates are always better for identifying which variables drive separation.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A data analyst has created a t-SNE plot of a high-dimensional dataset and a scatterplot matrix of the original variables. The analyst wants to select a cluster of points in the t-SNE plot and see where those same points fall in the scatterplot matrix. Which interactive technique should the analyst use?

   • AZooming into the t-SNE plot.
   • BBrushing and linking between the two views.
   • CSorting the scatterplot matrix by the t-SNE coordinates.
   • DChanging the color palette of the t-SNE plot.
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   What does 'linking' mean in interactive multivariate visualization?

   • ACombining multiple datasets into one for visualization.
   • BSynchronizing selections and highlights across multiple views of the same data.
   • CCreating hyperlinks to external documentation.
   • DConnecting data points with lines to show their order.
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   A data analyst has a dataset with 20 numeric variables and wants to identify groups of variables that are highly correlated with each other. The analyst has already computed the correlation matrix. Which approach is most effective for revealing these groups?

   • AApply PCA and plot the loadings of the first two components.
   • BCreate a scatterplot matrix of all 20 variables.
   • CCreate a parallel coordinates plot of all 20 variables.
   • DCreate a heatmap of the correlation matrix with hierarchical clustering applied to reorder rows and columns.
   Reveal Solution Discussion Select an answer first