Arcin, Alan ChesterGuliman, Ma. Ellysah JoyCenteno, Genna PaolaHerbo, Jacqueline MargauxParmanand, SanjeevMapa, Cherrie2024-09-202024-09-202024-07-169788413962016https://riunet.upv.es/handle/10251/208401[EN] Big data and machine learning applications are increasingly gaining traction in central bank operations. Among8 others, central banks have tapped big data in their nowcasting exercises. In this study, we construct inflation news indices and examine if such indices can help predict inflation. The indices are developed using lexicon-based sentiment analysis refined using reinforcement learning and supervised machine learning methods, particularly artificial neural networks and long short-term memory models. These indices are then used as additional feature variables in time series models and machine learning models for nowcasting regional and nationwide inflation in the Philippines. We find empirical evidence that our constructed inflation news indices can improve the predictive capability of these forecasting models.8Reconocimiento - No comercial - Compartir igual (by-nc-sa)InflationNewsNowcastingSentiment analysisMachine learningCapítulo de libro10.4995/CARMA2024.2024.17441Abierto