A carregar...
Publicação
Nowcasting inflation expectations using twitter
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 3.67 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
This study explores the potential of Twitter to forecast inflation expectations through
various machine learning models. Twitter-based measures of inflation were shown to be
correlated with survey-based inflation expectations. No single regression model was
consistently superior, although PCA-Ridge appears to be appropriate. The study also unveiled
potential issues with data cleaning, model overfitting, and limitations in available data. These
findings advance the understanding of economic forecasting using unconventional data sources,
opening pathways for future research.
Descrição
Palavras-chave
Nowcasting Twitter Inflation expectations Dynamic topic modelling CEMS MIM