DATA11001 INTRODUCTION TO DATA SCIENCE EPISODE 4: ADVANCED TOPICS IN VISUALIZATION TODAY’S MENU

1. NETWORKS

2. SPATIO- TEMPORAL DATA

3. INTERACTIVE VISUALIZATION NETWORKS/GRAPHS

• Topological structures, such as orderings, hierarchies (trees), and networks require visualization techniques different from those for metric data (1D or 2D plots, most notably)

• Plotting graphs can suggest more information than there is:

= NETWORKS/GRAPHS

• Topological structures, such as orderings, hierarchies (trees), and networks require visualization techniques different from those for metric data (1D or 2D plots, most notably)

• Plotting graphs can suggest more information than there is:

= NETWORKS/GRAPHS

• Topological structures, such as orderings, hierarchies (trees), and networks require visualization techniques different from those for metric data (1D or 2D plots, most notably)

• Plotting graphs can suggest more information than there is:

= NETWORKS/GRAPHS

• Data formats: – jsongraph

{ "graph": { "nodes": [ { "id": "A", }, { "id": "B", } ], "edges": [ { "source": "A", "target": "B" } ] } } NETWORKS/GRAPHS

• Data formats: – jsongraph – DOT () digraph G { A -> B; } NETWORKS/GRAPHS

• Data formats: – jsongraph – DOT (GraphViz) – GraphML (XML-based)

NETWORKS/GRAPHS

• Data formats: – jsongraph – DOT (GraphViz) – GraphML (XML-based) – CSV ()

A B # list no nodes 0 1 # adjacency matrix 0 0 # (asymmetric if directed graph) NETWORKS/GRAPHS

• Data formats: – jsongraph – DOT (GraphViz) – GraphML (XML-based) – CSV (adjacency matrix)

• Tools for handling and plotting graphs: – networkx (python) – GraphViz (for drawing) – network (R) – Cytoscape[.js] (originally for biological networks) TODAY’S MENU

1. NETWORKS

2. SPATIO- TEMPORAL DATA

3. INTERACTIVE VISUALIZATION GEOSPATIAL DATA

• The Flat Earth Society: making is easy GEOSPATIAL DATA

• GIS data can be represented as: – vector format: (points, lines, polygons) – raster format: (bitmap laid over a square area)

• Various formats: – Shapefile – GeoJSON – Keyhole Markup Language (XML-based)

• Tools: – python packages: shapefile, cartopy, ... – Google Maps API LAYERS LAYERS THE GOOD, THE BAD, AND THE UGLY #9: HOW TO LIE WITH STATISTICAL GRAPHICS: A LESSON BY NASA

• Summer 2017: Dramatic "footage" from a Pluto flyover: THE GOOD, THE BAD, AND THE UGLY #9: HOW TO LIE WITH STATISTICAL GRAPHICS: A LESSON BY NASA

Fundamental principle: Above all else show the data.

THE FINE PRINT TODAY’S MENU

1. NETWORKS

2. SPATIO- TEMPORAL DATA

3. INTERACTIVE VISUALIZATION INTERACTIVE VISUALIZATION

• Even very little interaction can help show a lot of information without9/9/2017 overloadinghttps://www.cs.helsinki.fi/u/ejaasaar/mrpt/results/stl10/Plot_100.html

STL-10, k = 100 1e+2 ann falconn flann-kd flann-kmeans kgraph rp trees sparse rp trees 2e+1 mrpt

1e+1

Trees: 1000.0 Depth: 13.0 Sparsity: 0.01

time 100 queries (s), Votes: 1.0 2e+0

1e+0

0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 recall None

https://www.cs.helsinki.fi/u/ejaasaar/mrpt/results/stl10/Plot_100.html 1/1 INTERACTIVE VISUALIZATION

• Interactive geospatial visualization INTERACTIVE VISUALIZATION

• Interactive geospatial visualization