Total retail consumer goods (TRCG) is an important indicator reflecting the macroeconomy, but due to its many influencing factors, it has a greater impact on forecasting accuracy. Therefore, accurate forecasting of total retail consumer goods is inseparable from effective screening and processing of indicators. By introducing the data of economic factors and power factors in the data preprocessing stage and determining the lag order of each influential factor, this paper realizes the purpose of exploring the future development trend of total retail consumer goods in advance. To identify the time-dependent relationship between the extracted features and the future total retail social consumer goods, the gated recurrent unit (GRU) model is constructed for forecasting. Finally, this paper empirically analyzes China's total retail social consumer goods from 2013 to 2019. The results show that compared with the benchmark model, the GRU has lower prediction error and higher prediction accuracy.