时间序列长度对基于Prophet的自动售货机销量预测影响研究

Research on the Influence of Time Series Length on Prophet-Based Vending Machine Sales Forecast

本次采用prophet模型对自动售卖机进行不同时间序列长度的销量预测,对预测效果进行对比分析,同时考虑外部天气因素对销量的影响。并且对自动售卖机的异常值采用基于离差系数和小波分析相结合的方法进行识别处理。实验结果表明,随着时间序列长度的增加,总体预测的精度逐渐提高;对于单机预测,需要具体问题具体分析,不同机器对于时间序列长度的敏感性不同。这为企业应对市场需求变化提供了重要的科学依据。

This time, the prophet model is used to forecast the sales of vending machines with different time series lengths, and the forecasting effects are compared and analyzed, and the influence of external weather factors on sales is considered. And the abnormal value of the vending machine is identified by a method based on the combination of dispersion coefficient and wavelet analysis. Experimental results show that as the length of the time series increases, the accuracy of the overall forecast gradually improves;for single-machine forecasting, specific problems need to be analyzed in detail, and different machines have different sensitivity to the length of the time series. This provides an important scientific basis for companies to respond to changes in market demand.