Cigarette market state monitoring based on retail terminal data governance

A set of monitoring methods for domestic cigarette market state based on retail terminal data governance was proposed to address the common problems of uneven quality, distorted QR code data and false inventory reporting in cigarette retail terminal data, which led to insufficient accuracy in the market state assessment. Cigarette market state monitoring methods based on retail terminal data governance were proposed and an integrated analysis framework of "data governance-state assessment" was built. At the data governance stage, firstly, cluster analysis was used to remove significantly abnormal terminals. Secondly, the maximum empirical likelihood density method was used to standardize the index scores. Thirdly, a latent factor model (LFM) was introduced to conduct a comprehensive evaluation on the data quality of retail terminals to select high-quality retail terminal samples. At the state evaluation stage, based on the selected high-quality terminals, a market state index system integrating dimensions of order sufficiency coverages, order sufficiency rates, gross profit margins, and salable days was constructed. After preprocessing the product specification data and calculating the composite scores, a Gaussian mixture model (GMM) was used to quantitatively divided the market state thresholds, and each product specification was mapped into five market status levels of "hot, tight, balanced, loose, and weak". The empirical analysis showed that this method effectively improved the accuracy of monitoring results and was highly consistent with the actual market conditions, providing a scientific basis for the formulation of precise cigarette marketing strategies.