BILSTM Seq-to-Seq with Multi-Head Attention for Consumer Price Index Forecasting

The Consumer Price Index (CPI) measures price changes in consumer goods and services, serving as a key indicator of inflation with significant implications for economic policy, market decisions, and social stability. Accurate CPI forecasting is crucial for effective economic management. Existing CPI forecasting methods lack precise modeling of temporal patterns and struggle to capture long-term trends. To address these shortcomings, this study proposes a sequence-to-sequence model integrating bidirectional Long Short-Term Memory (BILSTM) networks and a self-attention mechanism. This model enhances temporal dependency capture and supports multi-step predictions. Tested on real-world CPI data, it outperforms baseline neural network models, offering a more precise and robust solution for CPI forecasting.