Using Machine Learning and Sentiment Analysis in Official Statistics

This paper provides a framework for analyzing different dimensions of Official Statistics based on the sentiments conveyed from various sources such as policy statements, expectation surveys, survey of professional forecasters, etc. This study for investigating the dynamics of inflation in India is based on the sentiments conveyed from Daily Newspaper, Household's Inflation Expectation Survey, Survey of Professional Forecasters on Macroeconomic Indicators and Consumer Confidence Survey. The study uses applications such as Mathematica, etc. to apply different Machine Learning Models to learn the outcome of sentiments and provide a platform for determining the Inflation Trend. Further, an overview of Blockchain application for distributed and secure ML Models for Official Statistics is also discussed