A Data-Driven Gross Domestic Product Forecasting Model Based on Multi-Indicator Assessment
Xin Wu, Zhenyuan Zhang, Haotian Chang, Qi Huang
Abstract
Gross domestic product (GDP) is a general reference to comprehensive measure the level of a country or region’s economic development and diagnoses the health of economy. Traditional economic census-based methods for GDP forecasting are often expensive and resource-consuming, more importantly, economic census results lag significantly. This paper presents a data-driven GDP forecasting model that integrates multidimensional data from the aspects of electricity consumption, climate and human activities. Specifically, the model is built up based on the long-short-term-memory neural network with particle swarm optimization algorithm. The input multidimensional data are analyzed by correlation-based feature selection, and then filtered to five influencing factors. The experimental results show that these influencing factors are obviously related to economic development, at the same time, GDP can be well predicted based on the proposed model in a timely and relatively accurate manner.