A Comparative Study of Time Series Data Forecasting Using Machine Learning Based on Improved Grey Model

This paper proposes a time-series data prediction hybrid model suitable for groundwater level to further study its impact on new energy power generation through the close relationship between groundwater level and climate. The model is based on the improved grey model (IGM) and combined with other models such as backpropagation neural network (BP) and general vector machine (GVM), which are used to predict time series data. Based on the time series data of groundwater level, the prediction ability of IGM BP, IGM RBF, IGM support vector machine, and IGM-GVM hybrid model is studied, which proved the feasibility of the model. The statistical parameter values finally conclude that the IGM-GVM model outperforms other selected models in the dataset utilized in the paper. In addition, the mixed model has a better effect on the dataset than the corresponding single model.