Visualizing Relationships between Global Indicators

Visualizing Relationships between Global Indicators Prabath Gunawardane † , Jack Feng † , Suresh K. Lodha † , Ben Crow * , Brian Fulfrost * and James Davis † † Department of Computer Science, University of California Santa Cruz, CA of Sociology, University of California Santa Cruz, CA {prabath, txfeng, lodha}@soe.ucsc.edu, {bencrow, fulfrost}@ucsc.edu, davis@cs.ucsc.edu • There is a large amount of data collected across all countries annually, over a range of socio-economic indicators. For example * Department the World Development Indicators Database [1] has data that covers 225 countries and regions, spanning 40 years for more than 500 indicators. While having more information is definitely better, visualizing this becomes a harder problem! • Our project focuses on developing an interactive interface to select slices from this large volume of data and perform correlation analysis, in turn displaying the results on a correlation matrix and a pseudo-colored world map. • The objective of this visualization is to begin to investigate deeper questions regarding the relationship between various global indicators and countries. Indicators (i) (ii) Time Countries Figure 1: The 3-dimensional volume of indicator, country and year data, with (i) a vertical 2D slice highlighted which shows times series data for all indicators for a specific country, and (ii) a horizontal 2D slice show- ing time series data for a single indicator over all countries Figure 2: The main interface, which allows user to pick a query type and select a set of countries, indicators and a time period. (i) Multiple indicator correlation for a single country : We can study the correlation of multiple indicators over time for a specific country. This allows us to look at questions such as ‘From 1980 to 2000 in Malaysia, which socio-economic indicators were strongly linked together?’ GDP per capita Population Total Population Density Female Labor Force % Urban Population % Trade % Adjusted Net Savings Pick a single indicator (a column) and sort. Adult illiteracy rate Agriculture (% of GDP) Figure 3: Correlation matrix for 29 indicators in Malaysia, from 1980 to 2000. Each cell represents the correlation coeffi- cient C ij between the two indicators along the i th row and j th column, which can be identified using the interface. Figure 4: The correlation matrix sorted on ‘GDP per cap- ita’ column, making it easier to identify the indicators that correlate well or inversely with GDP. (ii) Indicator trends between countries : It may also be insightful to compare how a single indicator varied over a time, across a set of countries. For example, the changes in the female labor force from 1980 to 2000. Pick a single coun- try and visualize that column on a world map. Figure 5 : Correlation matrix for 207 countries, on female labor force (as % of total) from 1980 to 2000. While this matrix shows all the countries, it is easy to select only a subset of countries to re- duce the amount of data shown. References [1] World Development Indicators 2002, World Bank. Figure 6 : After picking out a single country (China), we can visualize the correlation of female labor force between the chosen country and other countries.