Data Visualization and Journalism Workshop: Many Eyes

Preview:

DESCRIPTION

Presentation slides from the 1st meeting of the Data Visualization and Journalism Working Group at UW-Madison J-school. Workshop on the tool Many Eyes.

Citation preview

DATA(VIS) JOURNALISM

-Working with Many Eyes

Nakho Kim (nkim3@wisc.edu)Dec 2011

Short Recap

• (Roughly) 4-step procedure– Plan: have a story to tell– Collect: choose data– Process: cleanup / analyze– Show: choose visuals

• Pattern finding – Difference – Clustering – Overlap – Change over time

Introducing Many Eyes

• Why– Easy to use: menu-based.– Somewhat interactive.– Good enough for 'familiar' vis work.– Good for trying out different types of vis on the spot.

• Limits– Not very flexible.

• Cannot designate own colors• Cannot use custom shapes• Cannot modify other’s vis. • Cannot…

(cont.)

• Runs as a Java applet– Good for cross-browsing but needs plugin.– Tries to become a kind of Youtube of datavis.

• Has been used for a variety of topics– http://

www.guardian.co.uk/news/datablog/2011/mar/17/visualise-data-trends

Datasets

• Register to use your own dataset.• Everything becomes public – So keep a source for your dataset.

• Upload– Open up as csv or spreadsheet, copy-paste your

data directly. – Check preview.

Data Cleanup

• The first row is the header. – Required. – Only one.

• Units should be put in labels.– Removed automatically from cells

• Use as little n/a as possible.• Remove sums.

Available VisFormat ME category Good for Scatterplot See relationships among data points ClusteringMatrix Chart See relationships among data points Difference Network Diagram See relationships among data points ClusteringBar Chart Compare a set of values Difference Block Histogram Compare a set of values Difference Bubble Chart Compare a set of values Difference Line Graph Track rises and falls over time Change over timeStack Graph Track rises and falls over time Change over timeStack Graph for Categories Track rises and falls over time Change over timePie Chart See the parts of a whole Difference Treemap See the parts of a whole Difference Treemap for Comparisons See the parts of a whole Difference Word Tree Analyze a text ClusteringTag Cloud Analyze a text ClusteringPhrase Net Analyze a text ClusteringWord Cloud Generator Analyze a text ClusteringWorld Map See the world Overlap (as of Dec 11)

Scatterplot

• Good for: correlation• Column Data Format– 1 Label – 3 values : x, y, size

Matrix Chart

• Good for: many comparisons at once– Draws a matrix of multiple bar or pie charts

• Format– 2 Labels (x, y) or 3 Labels (x,y, color)– 1+ values

Network Diagram

• Format: – 2 values: from, to.

• Better done w/ Gephi, NodeXL, TouchGraph.

Bar Chart

• Format– 1 label.– 1+value(s)

Block Histogram

• Good for: – Distribution– Interface – However, can't designate order

• Format– 1 label– 1 value

Bubble Chart

• Good for– Comparing many data at the same time.– But colors are random. – Use only for positive values. – Can use 2 levels (shown as pie).

• Format– 1-2 label– 1 value

Line Graph

• Always have series as columns (vertical). • Format: – 1 label– series 1, series 2...

Stack Graph

• Good:– Time-series change of ratio. – Use only positive values. – Y-axis shows sum.

• Stack Graph for Categories– Groups categories by colorset.

• Format– 2+ labels. – series 1, 2...

Pie Chart

• User can switch btw series. • Format– Label– series_value1. value2...

Treemap

• Good for– hierarchial allocation.

• Format– 1+ hierarchial label– value1 (area), value2(color).

• Treemap for Comparisons– 'change tree map'. change over time series.

Word Tree

• Parses out recurrent contexts. • Format: free text.

Tag Cloud

• Parses out recurrent keywords. • Format– 2 col data– Text– Or fragments.

• Word Cloud Generator– A variation

Phrase Net

• Networks words. designate filter word and draw graph.

• Format: text.

Maps

• Better done w/ Fusion Tables• limited regions. • But can have 2 or all comp. colors or bubbles.

can't overlap multiple values. • Format– region(name, ISO ccode, standard abbr)– value (# or txt)

Examples of the Week

• Good– The Guardian: Twitter rumor spreading– http://www.guardian.co.uk/uk/interactive/2011/dec/07/london-riots-twitter

– Rich information interface– Vivid change over time, as expressed with time

• Good or Bad– Yonhap: Credit level change over time – http://www.yonhapnews.co.kr/medialabs/sound/credsound.html

– Decoding information from sound– Uses for the visually impaired?

Recommended