Data Vizualization
Univariate vizualizations


•Univariete = only one variable is visualized
•
•Different ways of visualizations for different data types
•Categorical – comparing categories (bar chart)
•Fraction – value relative to the whole (pie charts)
•Cardinal – distribution = „spread of data“ (histogram, box plot)

Pie chart
•Only small number of categories (max about 7)
•Not very informative when all slices are of similar size
•Only when it is part of whole




Making of pie chart
•Counts for categories are needed
•







Bar chart
•Useful for almost any situation
•Base for numerous other charts
•Very easy for interpretation
•Unlimited number of categories
•(then it is turned into histogram)
•




If we want percentages, we have to calculate them






Stacked bar plot
•


Stacked bar plot



•



Histogram



•Avoid the usage of 3D versions
•They can be very misleading
•Always make titles, subtitles and labels as parsimonous as possible
•(parsimony means to be maximally simplistics and maximally informative simultaneously)
•