Summary Data analytics for engineers

EXPLORATORY DATA ANALYSIS  Shapes
Key features:
 Getting to know the data Channels (appearance of marks):
 Extensively using graphs  Position
 Generating questions  Color
 Detecting errors in data  Length
 Size
DATA TYPES  Orientation
Categorical/qualitative
 Dichotomous DISTRIBUTION SHAPES
 Yes/no, male/female
 Nominal (sometimes called categorical)
 Movie genres (no ordening)
 Ordinal (sometimes called ordered)
 Movie ratings, good/better/enz.
(ordening)

Numerical/quantitative
 Interval (no 0 point, only difference has
meaning) Right skewed: long tail to the right, mean bigger than
 World ranking, release date median, median closer to 25th quartile
 Ratio (0 or larger, meaningful to calculate with)
 Height, weight, traffic accidents PLOTS AND VISUALIZATIONS
SCATTERPLOT
VISUALIZATION Data type: 2 numerical
Key features: Key or value: no keys, only values
 Exploration, for you (finding patterns) Marks and channels: points, using positioned
 Confirmation (seeing patterns that match your horizontal and vertical
expectations) Usage:
 Communication, for others (simple, selected,  Shows structure (not for large datasets)
summarized)  Find trends, outliers, distribution, correlation
and clusters
TABLES OR GRAPHS?
Graphs:
 For making comparisons
 To get a quick view
 Use to discover relations

Tables:
 For reading of values
 To draw attention to sizes

Reference table: store all data to lookup data
Demonstration table: to illustrate a point so present BAR CHART
just enough data Data type: 1 categorical, 1 quantitative
Key or value: 1 key, 1 value
KEYS AND VALUES Marks and channels: lines, using length and spatial
Key attribute: regions
 Independent Usage:
 Categorical  Compare and lookup values

Value attribute:
 Dependent
 Numerical/ quantitative

MARKS AND CHANNELS
Marks:
 Points
 Lines
 Areas

, STACKED BAR CHART KERNEL DENSITY PLOT
Data type: 2 categorical, 1 quantitative Data type: -
Key or value: 2 keys, 1 value Key or value: -
Marks and channels: lines, using length, spatial Marks and channels: -
regions and hue Usage:
Usage:  No fixed bins so overcomes drawbacks of
 Compare and lookup values histograms
 Overview what the relation is to the whole  Explore distribution shape




LINE CHART
Data type: 2 quantitative, never use categorical
Key or value: 1 key, 1 value
Marks and channels: points with line connections, CUMMULATIVE HISTOGRAM
using aligned lengths Data type: -
Usage: Key or value: -
 Find trend and relationship between item and Marks and channels: -
next item Usage:
 Illustrate thresholds




HEATMAP
Data type: 2 categorical, 1 quantitative
Key or value: -
Marks and channels: area, using color BOX PLOT
Usage: Data type: summary statistics
 Find clusters and outliers Key or value: -
Marks and channels: -
Usage:
 To compare groups
 Find distributions (uses median, Q1, Q3, min
and max)




HISTOGRAM
Data type: numerical
Key or value: -
Marks and channels: -
Usage: VIOLIN PLOT
 Shows distribution shape Data type: numerical
Key or value:-
Marks and channels:-
Usage:
 Same as boxplot but also has kernel density
plot on each side which means they also show
the probability