Interactive federal statistical data on the Web using nViZn

1
Interactive federal statistical data on the Web
using nViZn

• Jon Hurst, Jürgen Symanzik,
• Lacey Gunter
• Utah State University
• email jon_at_jonathan.hurst.name
• Web site www.jonathan.hurst.name/academic

2
Contents

• Background Information
• Our software
• Summary

3
Background gt NSF Digital government Program

• Mission explore and develop new information
  technologies that will improve the way government
  serves the American people.
• Digital Government Quality Graphics initiative
  disseminate federal statistical data over the
  Internet in more usable and understandable forms.
• Resources http//www.diggov.org
• http//www.geovista.psu.edu/grants/dg-qg/intro.h
  tml

4
Background gt Micromaps

• Micromap is short for linked micromap plot.
• First presented in 1995 by Olsen, Carr, Courbois,
  and Pierson
• Help focus the viewers attention on the
  geographical significance of the data.
• Resources
• http//cm.bell-labs.com/cm/ms/who/cocteau/newslet
  ter/index.html

5
Example of a linked micromap plot
6
Background gt Interactive micromaps on the Internet

• Planned, but never used, for the EPA Cumulative
  Exposure Project (CEP)
• USDA-National Agricultural Statistics Service
  uses pseudo interactive micromaps
• http//www.nass.usda.gov/research/sumpant.htm
• The National Cancer Institute plans on using them
  in a web site to be made available this year.
  http//statecancerprofiles.cancer.gov/micromaps/

7
Our Software gt Overview

• Goals
• Easy access
• Concise display of the CEP data
• Understandable to a non-statistical audience
• Solutions
• Software accessible from the Internet
• nViZn SDK
• Interactive drilldown maps, micromaps, and tables

8
Our Software gt nViZn

• A commercial, Java-based SDK for visualization
  and analytics
• Successor to the GPL
• Based on Wilkinsons The Grammar of Graphics
• Includes libraries and sample code and
  applications
• Resources http//www.spss.com/nViZn/

9
Our Software gt Background

• Where possible, based on samples provided with
  nViZn
• Built on the previous code written by Lacey Jones
• Currently a prototype for an Internet application
  (such as would have been used by the EPA CEP)
• Final version?

10
Our Software gt Interface

• Drilldown map
• Interactive tables
• Interactive micromaps
• Meta displays
• Live demo ?
• (slides that follow are of images outlining the
  live demo)

11
Drilldown Map
12
Interactive tableMassachusetts benzene
concentrations sorted by minimum values.
13
Interactive tableNew York benzene concentrations
sorted by mean values.
14
MicromapNew Mexico benzene concentrations sorted
by median values.
15
MicromapCalifornia lead concentrations sorted by
mean values.
16
Comparisons between micromapsWashington Oregon
median acrylic acid values.
17
(No Transcript)
18
Our Software gt Implementation gt Data

• Data is a duplicate of the EPA CEP
  datahierarchical file structure
• 148 HAPs modeled for 60,000 census tracts
  according to 1990 data
• Organized by FIPS codes
• Outdated compared to databases
• ArcInfo gen files required for micromaps

19
Our Software gt Implementation gt nViZn

• nViZn libraries handle the data, analytics
• They also handle the creation of all graphics,
  and
• Interaction with the graphics
• A knowledge of Java is required
• Simple graphs are quick and easy
• Interactive applications are long and dirty

20
Our Software gt Implementation gt nViZn

• Code for using nViZn is usually concise
• The learning curve is steep

21
Our Software gt Implementation gt Summary

• About 2,000 lines of code
• About 200 hours for someone familiar with
  programming, but having never used Java or nViZn

22
Our Software gt Missing features

• Database connectivitynViZn libraries support
  database connections
• Applet enabledcurrently it is a stand-alone
  application

23
Summary gt nViZn

• Good communication with users
• Takes the dirty work out of graphics
• Very robust
• Huge potential for use with Federal statistical
  data, and especially geographically referenced
  data
• Training offered by SPSS

24
Summary gt nViZn

• Poor documentation
• In some instances, no documentation or samples
  for features
• Desperately needs a manual
• Development time was unnecessarily long due to
  the lack of information

25
Summary

• Usability study needed how do non-statisticians
  understand these displays