Improvement in Inflation Forecasting: Ensembling Text Mining with Macro Data in Machine Learning Models

Pijush Kanti Das and Prabir Kumar Das

International Journal of Economics and Finance, 2024, vol. 16, issue 6, 92

Abstract: We forecast inflation using a large news corpus and machine learning methods. Over 3.9 million daily newspaper headlines from January 2001 to June, 2023 are decomposed into monthly time series and integrated with machine learning models to predict inflation. The addition of Text mining in models outperformed the numerical predictions based on the machine learning models without text mining as published by the authors earlier in Das and Das (2024). In addition, the variable importance while analyzing the predictors provides further insights into new variables came out from text mining for which structured data was not available earlier. A dictionary of words sentimental to inflation forecasting has been prepared possibly for the first time. The forecasting model that used text words sentimental to inflation as additional inputs in artificial neural network performed better than all the other models in terms of forecast accuracy. Overall, we provide a novel representation of improvements in adding text mining in machine learning models in inflation forecasting.

Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://ccsenet.org/journal/index.php/ijef/article/download/0/0/50265/54408 (application/pdf)
https://ccsenet.org/journal/index.php/ijef/article/view/0/50265 (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:ibn:ijefaa:v:16:y:2024:i:6:p:92

Access Statistics for this article

More articles in International Journal of Economics and Finance from Canadian Center of Science and Education Contact information at EDIRC.
Bibliographic data for series maintained by Canadian Center of Science and Education ().