Based on the idea of a semi-parametric generalized additive model, the factors influencing the national total retail sales of consumer goods are examined. Firstly, grey relational analysis and univariate analysis were used to screen out two nonlinear factors and three linear factors, and interaction effects were found between the two nonlinear terms. Finally, a semi-parametric generalized additive model containing interaction terms was established, and R-squared of the model reaches 97.7%, indicating that the model has extremely high explanatory ability. The results are as follows. M1, GDP and per capita disposable income of rural households have positive linear effects on the total retail sales of consumer goods in China. The influence of M2 and national per capita disposable income is non-linear. When per capita disposable income is low, the larger M2 is, the lower the total retail sales of consumer goods.