Machine learning-based retail terminal cigarette commodity order quantity automatic prediction method

The invention discloses a retail terminal cigarette commodity order quantity automatic prediction method based on machine learning, and the method comprises the steps: constructing a basic sample set based on historical sales information, constructing behavior features through the basic sample set, and extracting data features from the basic sample set; integrating the basic sample set, the behavior features and the data features to obtain a feature set containing a plurality of feature vectors, and performing feature selection on the feature set to obtain historical sales analysis data of the user on tobacco; and performing multiple iterations on the prediction model based on historical sales analysis data to obtain a final prediction model, and predicting the order quantity of the cigarette commodities by using the final prediction model. According to the method and the device, the historical sales information of the tobacco by the user can be deeply mined, and the logic information implying the user behavi