Project Team
- Dr. Deepak Padmanabhan , Queen's University Belfast, UK [Principal Investigator]
- Dr. Amit Thorat , Jawaharlal Nehru University, India [Lead International Partner]
- Prof. Amaresh Dubey , Jawaharlal Nehru University, India
- Dr. M. Satish Kumar , Queen's University Belfast, UK
- Carl Duffy, Queen's University Belfast, UK
- Soumali Banerjee, Jawaharlal Nehru University, India
Project Details
- Funder: Department for the Economy, Northern Ireland, UK
- Budget: £19,124
- Duration: February - July, 2017
Project Publications
- Carl Duffy, Deepak P, Cheng Long, Satish Kumar, Amit Thorat, Amaresh Dubey, "Fast Identification of Interesting Spatial Regions with Applications in Human Development Research", Short Paper in 29th International Conference on Database and Expert Systems Applications (DEXA 2018), September 2018, Regensburg, Germany
Map Visualizations
As part of the project, we generated visualizations of inter-attribute relationships between attributes across various Indian states. We took a variety of attribute pairs, such as job industry and ever attended school (the latter is a binary attribute), and used Fisher Exact Test to see whether there are relationships between them. We were simply interested in the existence/non-existence of a relationship, and not on the direction of relationship. Given the limited size of the dataset, we were unable to get statistically significant results in many cases, and thus, just present the p-values in a plot. A low p-value, specifically p < 0.05 or p < 0.01, exemplifies a statistically significant relationship. We link some representative visualizations herein. It may be noted that the data for Andhra Pradesh was from the pre-2014 Andhra Pradesh, and given the creation of Telengana and usage of a map with modern boundary, Telengana is always shown greyed; however, the p-value for Andhra Pradesh holds for Telengana as well.
- Ever Failed or Repeated a Class AND Farm Work Hours Per Day Link
- Industry of Job AND Literacy Link
- Number of Jobs Done for Money in Last Year AND Literacy Link
- Government Job AND Ever Attended School Link
- English Ability AND Farm Work Hours Per Day Link
- Work Hours Per Day AND English Ability Link
- Ever Attended School AND Farm Work Hours Per Day Link
- Number of Jobs Done for Money in Last Year AND Ever Attended School Link
- Number of Jobs Done for Money in Last Year AND English Ability Link
- Industry of Job AND Ever Attended School Link
- Work Hours Per Day AND Ever Attended School Link
- Government Job AND Ever Attended College or Vocational School Link
Project Goals
- Overlay IHDS data on maps to enable heatmap-based analysis to understand the overarching trends in facets such as living standards, income and consumption data.
- Deploy data analysis/mining algorithms for identifying India-wide attribute correlations and use those to formulate questions that merit further attention on India’s human development.
- Building upon techniques for spatial outlier detection, develop methods to automatically identify geo-clusters that buck particular country-wide trends that have been identified earlier.