Joint Utilization of Geospatial and census proxies for Autoencoder-Assisted Downscaling (JUGAAD) of socioeconomic indicators in India

2026-07-17

AI Summary

This study addresses the spatiotemporal mismatch between census and survey data in developing countries by proposing a high-resolution downscaling method for socioeconomic indicators. The approach first aggregates census and geospatial data to village-cluster units to reduce noise, then employs an autoencoder to extract a low-dimensional latent representation from nationally representative survey data (NSSO). A regression model subsequently maps the aggregated data onto this latent space, enabling predictions at fine-scale census resolutions. By innovatively integrating heterogeneous, multi-source data, the method effectively mitigates challenges posed by administrative boundary changes and data sparsity while substantially improving prediction accuracy. Experiments using Indian data from 2001 and 2011 demonstrate strong agreement between the generated high-resolution indicators and district-level NSSO ground truth values.

Abstract

Monitoring poverty and food security indicators is imperative for addressing socioeconomic challenges in developing nations. A limitation is mismatches in scale between data sources: census data provide geographic coverage, while socioeconomic indicators are derived from infrequently conducted surveys at coarse resolutions, posing a methodological challenge. This study introduces a deep learning framework, JuGAAD, using Indian census and survey data from 2001 and 2011 as a case study. We employ a three-step process: census and geospatial data are averaged into intermediate village-cluster-scale tessellations to reduce noise and regularize administrative boundary changes; an autoencoder compresses high-dimensional National Sample Survey Office (NSSO) data into a low-dimensional latent representation; and a regression model maps upscaled census and geospatial data to this representation. This function is applied to fine-grained census data to generate high-resolution predictions, validated against ground-truth district-level NSSO indicators. Results confirm the methodology predicts socioeconomic indicators at fine scales with strong accuracy.

Problem

• poverty monitoring
• data scale mismatch
• socioeconomic indicators
• spatial downscaling
• census and survey data

Innovation

• autoencoder
• downscaling
• geospatial data
• census data
• socioeconomic indicators