In the environment that the passenger car industry is becoming increasingly prosperous, it is crucial for each automobile enterprise to accurately grasp the development direction of the industry and formulate suitable production goals. In this paper, historical sales, macroeconomic indicators and online search keyword data are selected as variables to establish a variety of sales prediction models for the overall passenger car market in order to improve the prediction accuracy of passenger car sales. Through comparative analysis, the Gradient Boosting Decision Tree (GDBT) algorithm model considering the above 3 variables has the best effect and its Mean Absolute Percentage Error (MAPE) is 10.35%. The model obtained in this paper can help automobile enterprises understand the development of market trends, make targeted production planning and provide a new reference model for the research of sales forecast.