Forecasting Economic Trends with Generative AI: A Python Framework for Policy Analysis

This paper develops a Python-based framework integrating generative AI with traditional economic analysis to enhance UK economic trend forecasting. Traditional approaches are not suitable to process large volumes of diverse data, particularly when combining structured economic indicators with unstructured policy documents and sentiment data. The methodology employs a mixed-methods approach using BM25 keyword retrieval, vector-based embeddings, and semantic hybrid retrieval. DeepSeek AI integration provides natural language processing capabilities, whilst ARIMA time series forecasting enables economic projections. Comparative evaluation demonstrates significant improvements over traditional systems. For UK unemployment analysis, the AI-enhanced approach delivered comprehensive insights identifying specific dimensions including post-pandemic impacts, health-related inactivity, and regional disparities, whilst traditional systems returned fragmented, limited-relevance documents. The framework contributes practical AI-economic analysis integration methodology, innovative policy document analysis through embedding-based contextual search, sentiment analysis integration in forecasting, and structured evaluation frameworks. Key limitations include temporal data lag, declining long-term forecast accuracy, API rate limitations, and UK-centric evaluation. The paper demonstrates AI's potential to complement human expertise in economic analysis, creating synergies that transform understanding, prediction, and response to economic challenges in data-rich environments.