数据驱动卷烟市场调控的探索与实践

“十二五”以来，卷烟消费市场快速变化，零售终端升级转型，品牌竞争日趋激烈，市场把控难度加大。本文通过分析宏观经济、批发、零售、消费和专卖管理等数据，用3σ法则确定零售库存的阈值区间，从总量、价位、品规等多维度综合判定市场状态，建立了以多层神经网络算法为核心的市场预测办法，进而运用预测结果开展市场状态调控，构建了数据驱动市场调控机制，有助于实现卷烟市场状态的“稍紧平衡”。 

Exploration and practice of data-driven cigarette market regulation

Since the "Twelfth Five-Year Plan", cigarette consumer market has changed rapidly, retail terminal has been upgraded and transformed, brand competition has become increasingly ferce, and market regulation has become increasingly difficult. By analyzing the macroeconomic, wholesale, retail, consumption and monopoly management data, this paper uses the 3σ rule to determine the threshold range of retail inventory, comprehensively determines the market status from total amount, price level and product regulation, and establishes a market forecasting method using multi-layer neural network algorithm. Based on the forecasting results, market status regulation and control is carried out, and a data-driven market regulation mechanism can be built, which helps to achieve a "slightly tight balance" in the cigarette market.