沃尔玛百货公司零售数据预测分析

Wal-Mart Retail Sales Data Forecast Analysis

本文从Kaggle网站上获取到沃尔玛在不同地区45家门店2010年、2011年、2012年的销售信息。首先,对数据进行纵向和横向统计分析,获取到门店的销售信息,并发现在2010-02-12、2010-09-10、2010-11-26、2010-12-31、2011-02-11、2011-09-09、2011-11-25、2011-12-30、2012-02-10、2012-09-07这些节假日的销售额会高于非节假日,对沃尔玛销售有益。接着,对解释变量进行相关性分析,发现燃油价的变动对居民消费指数影响较大,从而间接影响失业率,燃油价格升高在一定程度上会促进居民消费指数的上升,降低失业率,而燃油价格下降具有相反的变化趋势。然后,对数据进行标准化处理,使用多元线性回归模型在测试集上预测了销售额数据,并展示了部分预测结果及相关信息。最后,本文搭建了一个两层的深度神经网络模型并使用沃尔玛数据进行训练,该模型可用于预测其他门店的销售额。

This article obtained the 2010, 2011 and 2012 sales information of 45 Wal-Mart stores in different regions from the Kaggle website. First, the data is analyzed vertically and horizontally to obtain the sales information of stores. Found 2010-02-12, 2010-09-10, 2010-11-26, 2010-12-31, 2011-02-11, 2011-09-09, 2011-11-25, 2011-12-30, 2012-02-10, 2012-09-07 these holidays are sold would be higher than a non-holiday, which would benefit Wal-Mart sales. Then, the correlation analysis of ex-planatory variables found that the change of fuel oil price has a great impact on the consumer index, thus indirectly affecting the unemployment rate. The increase of fuel price will promote the in-crease of consumer index and reduce the unemployment rate to some extent, while the decrease of fuel price has the opposite trend. Then, the data were standardized, the multiple linear regression model was used to predict the sales data on the test set, and some of the predicted results and cor-relation information were displayed. Finally, in this paper, a two-layer deep neural network model is constructed and trained using Wal-Mart data. This model can be used to predict sales of other stores.