Measuring the international digital divide: an application of Kohonen self-organising maps

Abstract

Abstract:With the help of a Kohonen self-organising algorithm, this paper presents a mapping and analysis of the global digital divide along with its main drivers. Several broad groups and subgroups are identified, consisting of countries that are similar in their digital development and in a number of other attributes. We find that the digital divide seems to occur synchronously with divisions in income, social, demographic and infrastructure measures. By examining a large dataset of 160 countries over a short period of three years, we find evidence of both convergence and divergence among the countries over time. We expect these findings to inform the ongoing debate on drivers of the International Digital Divide (IDD). In addition, this paper provides a novel visualisation of the digital divide and its predictors on a two-dimensional grid. Extensions of this work, with the availability of more years of data, could investigate the potential convergence of countries to particular patterns of digital development.

Positioning Genomics in Biology Education: Content Mapping of Undergraduate Biology Textbooks

“…Clusters were identified using Kohonen Self-Organizing Maps (SOM), an application of a competitive neural network algorithm ( 16 , 18 , 19 ). Algorithms and applications of SOM methods have been described elsewhere ( 9 , 17 ). Analysis was performed using a standard statistical package, MATLAB 6.1.…”

Exploring Patterns in ICT Growth and Development in Africa: A Kohonen Map Analysis

“…This paper thus attempts to fill that gap by employing the SOM algorithm on data from 48 African countries covering a period of seven years (2001)(2002)(2003)(2004)(2005)(2006)(2007) to address the question of just how much ICT has evolved over this time period and the associated patterns in economic, social, and cultural factors. Deichmann et al (2007) and Hua, Skaletsky, and Westermann (2009) pointed out in their studies that Kohonen maps can be compared to a factor analysis followed by a cluster analysis and are a very effective tool for clustering data. Unlike the SOM property, factor-cluster analyses do not allow for clear visualization of results.…”

“…A major advantage of Kohonen maps is the self-organizing property of the map which makes estimated components vary in a monotonic way across the map (Deichmann et al 2007). This allows us to easily identify the characteristics of each cluster on the map; for example, moving from the bottom to the top of the map implies an improvement in digital development.…”

“…Hua, Skaletsky and Westermann (2009) also utilized SOM to evaluate the Central Intelligence Agency (CIA) World Factbook data and to identify the existence of any patterns among selected countries vis-à-vis their people, economies, communications, and defense forces. Similarly, Nguyen, Haighton, and Hudson (2008) and Deichmann et al (2007) used SOM to evaluate the levels of living standards among Vietnamese provinces, and the relative evolution and determinants of international digital divide, respectively. Consistent with similar studies, this paper utilizes the SOM technique in the analysis of our data.…”

Cities and Agglomeration Externalities

“…Most Chinese cities are undersized; migration restrictions led to loss in net output per worker of 17 per cent. For at least a quarter of the cities, these losses may range between 25 and 70 per cent World Bank (2009) Firms in more populated cities of China tend to be more productive-based on survey of 12,400 firms in 120 cities Da Mata et al 2007Increases in size of local markets and access to domestic markets strongly impact on growth rates of Brazilian cities-based on aggregate data for 123 cities for 3 decades from 1970 Deichmann et al (2007) strong influences of existing firms in cities and of localization economies are evident in plant location choices in Indonesia shukla (1984) Resources in an Indian city with a population of 100,000 can be, on an average, 23 per cent more efficient than in one of 10,000. They can be as much as 51 per cent more efficient in a city of one million than in a city of 10,000 Hansen (1990) Increasing distance from the city centre by 1 per cent leads to a 0.13 per cent decline in productivity in Brazil-based on 1980 data for 356 new manufacturing firms Henderson (1994) Doubling of distance to a regional market centre lowers profits by 6 per cent-based on firm data in auto component and agricultural machinery in Brazil and the United states Mitra (2000) Positive urbanization economies exist in 11 out of 17 two-digit level industries in India-based on panel data for 15 major states Lall et al (2003) Urbanization economies are strong in manufacturing-based on estimation of separate cost elasticities with respect to four measures of agglomeration for eight industrial groupings and three size classes of plants, using micro data from Indian survey of Industries 1998 Combes et al (2013) Even after controlling for individual and firms characteristics, the estimated elasticity of wages with respect to employment density in China is about three times larger than in Western countries.…”