Hybrid AI System Predicts Bolivia's Economic Roadblocks
A new hybrid probabilistic forecasting system has been created by researchers to anticipate roadblocks in Bolivia, a recurring issue that leads to economic losses estimated at 4% of the nation's GDP. This system, outlined in a paper on arXiv (2607.21785), combines time series decomposition (Prophet) with natural language processing (NLP) techniques, utilizing a six-year dataset of Bolivian news. It employs vector semantic embeddings and zero-shot classification models to identify signals of escalating discourse prior to the onset of roadblocks. Validation was performed using an expanding walk-forward method over 1,762 days, assessing seven forecasting horizons (H+1 to H+7) and comparing seven internal configurations alongside four external benchmarks, including SARIMA and LightGBM. The findings highlight the system's effectiveness for logistical planning in response to these disruptive occurrences.
Key facts
- Roadblocks cause losses equivalent to 4% of Bolivia's GDP.
- System uses Prophet for time series decomposition.
- NLP techniques applied to six-year Bolivian news corpus.
- Uses vector semantic embeddings and zero-shot classification.
- Validated over 1,762 days with seven forecasting horizons (H+1 to H+7).
- Compared seven internal configurations and four external benchmarks.
- External benchmarks include SARIMA and LightGBM.
- Paper published on arXiv with ID 2607.21785.
Entities
Institutions
- arXiv
Locations
- Bolivia