A Method for Predicting the Average Selling Price of Cigarette Per Carton Based on Multiple Linear Regression

As the cigarette industry shifted from the dual-growth model of "sales + structure" to the single-channel development model of "structure", the sustainable growth of the single-carton sales structure has become an important hand for the industry to realize high-quality development. However, the current prediction methods for the average price of cigarette sales per unit rely heavily on empirical forecasting, lacking in theoretical and scientific basis, and fall short of practical needs. In order to further improve the prediction accuracy of the average price of cigarette sales per unit, this paper combines macroeconomic indicators, constructs a multiple linear regression model, and screens the economic indicators and tests the model through the unit root testing, difference handing and autocorrelation test. Focusing on the cigarette business budget of Liangshan Yi Autonomous Prefecture, this study finds that the structure of cigarette sales per unit directly affects the average sales price, which is closely tied to consumer demand. It further reveals that cigarette consumption is influenced by seven macroeconomic factors: the growth rate of tax benefits, urban disposable income growth rate, national per-unit sales, GDP, the added value of the tertiary sector, CPI index, and household food expenditure. Consequently, a multivariate linear regression model is constructed with these seven factors as independent variables to predict the average price of cigarette sales per unit. The results demonstrate that this model can effectively explain 95.6% of the variance in the average price per unit, which can significantly enhance the accuracy and scientific rigor of measurable and controllable cigarette marketing with an average prediction accuracy of 99.58%.