Developing news-based Economic Policy Uncertainty index with unsupervised machine learning

Azqueta Gavaldon, A. (2017) Developing news-based Economic Policy Uncertainty index with unsupervised machine learning. [Data Collection]

Collection description

The EPU1 file shows the number of articles with the proportion of articles containing the list of keywords specified for each month and newspaper. For each newspaper, three components are shown, total number of articles that fulfil the set of keywords, total number of articles containing the word “today” and the proportion.

The EPU2 file shows the evolution of the 30 topics unveiled by the unsupervised machine learning algorithm. Details regarding the algorithm can be seen in section and three of the paper. Each topic has three components: the number of articles that represent the topic per month, the total number of articles containing the word “today” (proxy for the total number of articles and equal across all topics), and the proportion of articles describing each topic (articles describing each topic divided by the total number of articles).

The comparison file shows the evolution of EPU built with the two methods.

College / School: College of Social Sciences > Adam Smith Business School > Economics
Date Deposited: 28 Jun 2017 15:33
URI: https://researchdata.gla.ac.uk/id/eprint/433

Available Files

Data

Read me

Repository Staff Only: Update this record

Azqueta Gavaldon, A. (2017); Developing news-based Economic Policy Uncertainty index with unsupervised machine learning

University of Glasgow

DOI: 10.5525/gla.researchdata.433

Retrieved: 2024-03-28

Downloads

Downloads per month over past year