Scholarlink

This study aims at economic prediction in SAARC nations through the incorporation of Machine Learning (ML) and natural Language Processing (NLP) methods. The research takes a hybrid approach that incorporates structured economic statistics like inflation rates, GDP growth, and unemployment rates as well as unstructured data based on regional and global news articles’ sentiment analysis. Sentiment scores were calculated with proven tools such as VADER and TextBlob for measuring media and public perception. ML models such as Logistic Regression and Gradient Boosting were trained to detect the possibility of economic recessions and upcoming trends. For forecasting time- series, ARIMA and Simple Exponential Smoothing (SES) models were used, with ARIMA proving to be more accurate in most cases. The results showed that the countries with lower economic stability, like Afghanistan and the Maldives, posed more forecasting challenges. Also, an interactive Tableau dashboard was created to present main patterns and comparative analyses. In general, the research highlights the importance of joining both structured and unstructured data sources to enhance the quality and trustworthiness of economic forecasting.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Economic Stability in SAARC: An ML and NLP Approach to Forecasting and Crisis Prediction

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Tanya Das,
  • Anavi Jhunjhunwala,
  • Jasraj Singh Chopra,
  • Nischay Upadhyay,
  • Suresh B. Pathare

摘要

This study aims at economic prediction in SAARC nations through the incorporation of Machine Learning (ML) and natural Language Processing (NLP) methods. The research takes a hybrid approach that incorporates structured economic statistics like inflation rates, GDP growth, and unemployment rates as well as unstructured data based on regional and global news articles’ sentiment analysis. Sentiment scores were calculated with proven tools such as VADER and TextBlob for measuring media and public perception. ML models such as Logistic Regression and Gradient Boosting were trained to detect the possibility of economic recessions and upcoming trends. For forecasting time- series, ARIMA and Simple Exponential Smoothing (SES) models were used, with ARIMA proving to be more accurate in most cases. The results showed that the countries with lower economic stability, like Afghanistan and the Maldives, posed more forecasting challenges. Also, an interactive Tableau dashboard was created to present main patterns and comparative analyses. In general, the research highlights the importance of joining both structured and unstructured data sources to enhance the quality and trustworthiness of economic forecasting.