Network analysis for input output tables

Background

An input-output table records the transactions among the business sectors in an economy. It defines a special kind of weighted directed network. Regional, national or multi-national input-out tables are challenging to produce, but in recent years, many have become publicly available. Such tables are important subjects to study for economists, environmentalists, and government officials, among others, where advanced social network analytics have become powerful tools for providing novel insights into these tables as well as for uncovering their distinct properties.

Related work

There are two R packages that handle input-output tables. Package iotables provides pre-processing and basic analytical tools related to working with Eurostat's symmetric input-output tables as well as basic input-output economics calculations. Package ioanalysis provides fundamental input-output matrix calculations, rankings, coefficient analysis, impact analysis, and other table operators. The analyses from the two packages are mostly from the traditional economic perspective instead of the social network perspective.

Many network analysis packages are available. Some are comprehensive such as igraph, network, statnet, and sna. Some focus on special problems such as fastnet for larget network analysis and qgraph. Some specialize on certain network measures; e.g., packages centiserve, CINNA, influenceR, keyplayer, and ranknetr all focus on network centrality measures. Given the special structure of input-output networks, the generic network analysis tools could be customized with recent advances in methodology.

Details of your coding project

This project aims to develop an R package ionet with network functionalities specialized for those generated from input-output tables. The function modules include

• access interfaces to publicly available input-output databases;
• conversion the tables to weighted directed network data structure supported by igraph;
• streamlined network analysis demonstrations;
• new network methodologies for input-output tables (e.g., centrality measures with prior information).

Expected impact

The package will benefit researchers working on input-output tables as well as researchers with working on network analysis.

Mentors

• EVALUATING MENTOR: Jun Yan jun.yan@uconn.edu is a co-developer of R packages copula, geepack, and splines2, among others. He was a GSOC-R mentor in 2018.
• Panpan Zhang panpan.zhang@pennmedicine.upenn.edu is an expert in network analysis and applications. He is a co-developer of the R package wdnet.

Tests

Contributors, please do one or more of the following tests before contacting the mentors above.

• Easy: install packages iotables and ioanalysis; take one table from iotables as an example and run the economic analysis available in ioanalysis; install package igraph and use it to obtain the centrality measures for the network from this example table.
• Medium: set up a GitHub repo with a new R package; use r-lib/actions to automatically check the package after every commit; play with the GitHub action so that some test functions are run.
• Hard: implement Dijkstra's algorithm with C++ with an Rcpp interface in your package. Run the test on the Wiki page.

Solutions of tests

We have identified a student contributor. Please consider other projects. Contributors, please post a link to your test results here.

• EXAMPLE CONTRIBUTOR 1 NAME, LINK TO GITHUB PROFILE, LINK TO TEST RESULTS.
• Shiying Xiao, GitHub Profile, Easy Test, Medium Test, Hard Test.
• Fabrizio Sandri, GitHub profile, Easy tests and Medium-Hard tests
• Mechkak youcef amine GitHub Profile, Easy Test, Medium-Hard test