基于神经网络的网上商店商品排版分析

Analysis of Online Store Produce Typesetting Based on Neural Network

网上购物越来越主流化,带来网商的盛行。关于如何进行商品的排版,来使得网商获取到更大的利益,成为一个重要的问题。通常情况下,一种商品的实际获利,不仅仅是出价、进价还有销售数量的简单运算,可能还有其他的一些因素的干扰。基于此,提出一种基于神经网络的方式来对商品利润进行分析。主要依据商品的利润来进行商品的划分,以商品的利润为决策因素,根据以往的数据来分析。首先通过引入粗糙集理论,对影响商品利润的条件属性进行约简,获得约简后的属性及其相关的重要性;其次采用基于属性重要性的加权欧氏距离对数据分析,建立各个聚类的相关预测模型,并提取具有较高相似性的数据作为训练样本。以某超市的供销数据为例进行实验,结果具有一定的参考价值。

With the development of online shopping, online store is getting more and more popular. To gain greater benefits, how to typesetting the product grow to be an important question. As is known, the actual profit of a product is not only the bid and the purchase price, but also the simple calculation of the sales quantity, and there may be other factors. So, proposes a new way based on neural network to analyze the profitability of commodities. The goods are divided according to the profit. Takes the profit of the product as the decision factor and analyzes it based on past data. Firstly, uses the fuzzy rough set to reduce the attributes of many kinds of products that affect the commodity, and then,obtains the reduced attributes and their related importance. Secondly, uses the weighted Euclidean distance that based on the importance of attribute to analyze the data, then, establishes the related prediction models for each cluster, meanwhile, achieves the extract data with higher similarity as the training sample. Chooses the data of a supermarket, and the result can show something.