SHOW ME THE NUMBERS Designing your own data visualizations

September 2017

What is a tool?

How do we design an effective tool?

How do we design an effective tool?

Data Viz Principles

Program Context

Software Skills

Structured Dataset

Guiding Question

What can we learn from a car dashboard?

How many ways can you represent….

75 37 and

A few examples…

htt

p:/

/blo

g.vi

sual

.ly/4

5-w

ays-

to-c

om

mu

nic

ate

-tw

o-q

uan

titi

es/

Global

Single IM

Broad Focus Narrow Focus

PEPFAR Executives

TWG Leads

SI Advisers

Partners

Country Teams

Agency Leadership

Defining an audience

Sco

tt J

acks

on

, PW

C

Explore

Analyze

Communicate

Influence

A good visualization allows you to…

Ch

afet

z, E

ssam

, Hu

ghes

, Jo

hn

son

(2

01

6).

“Fu

nd

amen

tals

of

Dat

a A

nal

ysis

& V

isu

aliz

atio

n T

rain

ing”

Explore

Analyze

Communicate

Influence

Extract

Clean

Validate

….but first you need to start by setting up the data

Extract

Clean

Validate

Explore

Analyze

Communicate

Influence

Tool Building Process

Where do you look for data?

We are interested in tracking the progress of the number of people on

treatment by age group against the COP target set in PEPFARlandia.

• Where do we go to access and extract the data?

• What information do we need to pull the data?

Extract

Clean

Validate

Explore

Analyze

Communicate

Influence

Tool Building Process

Structure

Integration

Subset

Differing structures even within PEPFAR data systems

Extract

Clean

Validate

Explore

Analyze

Communicate

Influence

Tool Building Process

80%

How complete is the dataset?

775

777

745

75

779

785

786

Total Numerator

Aggregated Age/Sex

Results

Age/Sex Aggregated/Result

Age/Sex/Result

ServiceDeliveryPoint

ServiceDeliveryPoint/Result

PEPFARlandia Completeness Check FY 2016 Q2 HTS_TST

thousands of people

Validate the completeness of TX_CURR for DSD for the current FY in PEPFARlandia.

• Where do we go to access and extract/view the data?

• What information do we need to pull the data?

• What disaggregates do we need to look at?

How valid is the tool?

Panorama Tool

Investigate Adjust Tool Re-Validate

Extract

Clean

Validate

Explore

Analyze

Communicate

Influence

Tool Building Process

We are interested in tracking the progress PEPFARlandia is making.

• Using the handouts containing PEPFARlandia indicator data, explore the data by making some visualizations.

• What story are you trying to tell with the visualization?

Extract

Clean

Validate

Explore

Analyze

Communicate

Influence

Tool Building Process

Visual Analysis to Find Stories

An

dy

Kir

k, V

isu

aliz

atio

n W

ork

flo

w

o Categorical comparison and proportions

o Ranking: big, medium, small

o Measurements/values: absolutes

o Range and distribution o Context: Targets,

forecasts, averages o Hierarchical relationships

COMPARISONS

o Up and down vs flat? o Linear vs exponential o Steady vs fluctuating o Seasonal vs random o Rate of change vs

steepness

TRENDS

o Outliers o Intersections o Correlations o Connections o Clusters o Associations o Gaps

RELATIONSHIPS

We are interested in tracking the progress PEPFARlandia is making.

• How were you going about analyzing the data your visualizations?

• How could you improve upon your visualization?

Tools for Analysis

Extract

Clean

Validate

Explore

Analyze

Communicate

Influence

Tool Building Process

Static

Dynamic

Explanatory Exploratory

Form vs. Function

Static/Explanatory

htt

ps:

//w

ww

.was

hin

gto

np

ost

.co

m/g

rap

hic

s/lo

cal/

met

ro-c

losu

res/

Dynamic/Explanatory

htt

p:/

/flo

win

gdat

a.co

m/2

01

6/0

5/1

0/w

ho

-is-

old

er-a

nd

-yo

un

ger-

than

-yo

u/k

Static/Exploratory

htt

p:/

/ww

w.w

sj.c

om

/art

icle

s/u

-s-e

mp

loye

rs-a

dd

-27

1-0

00

-jo

bs-

14

46

81

67

37

Dynamic/Exploratory

htt

p:/

/ww

w.n

ytim

es.c

om

/in

tera

ctiv

e/2

01

5/1

1/2

4/u

psh

ot/

than

ksgi

vin

g-fl

igh

t-p

atte

rns.

htm

l

Consistent color scheme

Consistent font family, vary size and emphasis t T Adjust size

Simplify

Logical Order

Integrate text and legend MWI

TNZ

Start with a sketch

Key Principles

Clear purpose

We are interested in tracking the progress PEPFARlandia is making.

• How could you improve upon your visualization?

• Add in some features mentioned on the last slide around key principles, starting with a clear title

Extract

Clean

Validate

Explore

Analyze

Communicate

Influence

Tool Building Process

Extract

Clean

Validate

Explore

Analyze

Communicate

Influence

Tool Building Process

Let’s critique some data

visualizations

• Prepared for the ICPI/CIT training, “Data Use Training” (DC), October 2016 (adapted from June 2016 version)

• Presentation material adopted from J. Schwabish (2014). “A Visualization Mapping: Form and Function” (http://policyviz.com/a-visualization-mapping-form-and-function/), S.Ortiz (2012). “45 Ways to Communicate Two Quantities, (http://blog.visual.ly/45-ways-to-communicate-two-quantities), A. Kirk (2013). Visualization WorkflowFinding Stories and Telling Stories (http://www.slideshare.net/visualisingdata/andy-kirks-facebook-talk) and Chafetz, Essam, Hughes, Johnson (2016). “Fundamentials of Data Analysis & Visualization Training” (http://geocenter.github.io/StataTraining )

• Image Sources • Youtube [Dashboard], Sears Auto Center • Unsplash, [Lake and Mountains], Justin Luebke • Extreme Presentation, “Chart Suggestions” • Icons downloaded from the Noun Project and designed by various artists: The Noun Project, [map],

Ivan Colic; “Microscope”, lastspark, “Graph”, gira Park; “Influencer”, Adam Beasley; “Pickaxe”, Creative Stall, “Detergent”, Megan Mitchell, “Washing Machine”, National Parks Service, “Clipboard”, matthew hall, Phil Goodwin; [country shape], Anna Gajowiec, “Traffic Cone”, Vicons Design, “Structure”, Alexandr Cherkinshy, “simplify”, Chameleon Design, “Analytics, Syafiqa Fickle; “Pencil”, David; “Target, Franck Junker, “Mouse”, John Testa, “Wrench”, Bluetip Desig, “Power Drill” Maksim Karalevich

Notes and Attribution