Non-Ferrous Metal Price Point and Interval Prediction Based on Variational Mode Decomposition and Optimized LSTM Network

The accurate forecasting of metal prices is great importance to industrial producers as the supply raw materials a very important part production. futures market subject many factors, and are highly volatile. In past, most relevant research has focused only on deterministic point forecasting, with less performed interval uncertainty forecasting. Therefore, this paper proposes novel model that combines First, hybrid price was established using Variational Modal Decomposition (VMD) Long Short-Term Memory Neural Network (LSTM) based Sparrow Search Algorithm (SSA) optimization. Then, five distribution functions optimization algorithm were used fit time series data patterns analyze characteristics, Finally, optimal function results, range confidence level set determine model. validated by inputting copper aluminum into obtaining results. validation results show proposed not outperforms other comparative models in terms accuracy, but also better performance sharp fluctuations peaks, which can provide more valuable reference for investors.