Streamlining business functions in official statistical production with Machine Learning

TL;DR

This paper discusses the application of machine learning models to improve efficiency, accuracy, and responsiveness in official statistical production, based on pilot projects with real survey data from Spain.

Contribution

It introduces a practical approach to integrating statistical learning models into official statistical processes, emphasizing quality and efficiency improvements.

Findings

Enhanced accuracy in statistical outputs

Reduced response burden on survey participants

Improved timeliness and cost-efficiency

Abstract

We provide a description of pilot and production experiences to streamline some business functions in the official statistical production process using statistical learning models. Our approach is quality-oriented searching for an improvement on accuracy, cost-efficiency, timeliness, granularity, response burden reduction, and frequency. Pilot experiences have been conducted with data from real surveys in Statistics Spain (INE).