基于卷积神经网络和LSTM循环神经网络的客户复购预测方法

A prediction method of clients’repurchase based on CNN and LSTM RNN

通常一个销售代表会有数百名客户。销售代表无法定量预测哪位客户最近有下单需求,所以多采用轮询或者主观直觉的方式决定每天的回访客户名单。本文以深度学习的思路,把销售代表的历史回访记录作为输入数据,以卷积神经网络(CNN)和循环神经网络(RNN)为基础模型,让模型学习客户复购行为的内在逻辑以便指导初级销售代表的每日回访工作。相对传统回访策略,在深度学习算法辅助下的回访策略可以把回访相对成功率提高120%,取得了显著的效果。

Each sales representative usually has hundreds of customers.Sales representatives cannot quantitatively predict which customer will place an order recently,so they often use polling or intuitive methods to decide the list of visiting customers every day.This paper takes the historical visit records of sales representatives as input data,and uses convolutional neural network(CNN)and recurrent neural network(RNN)to make the model learn the internal logic of customer repurchase behavior,in order to guide the daily jobs of junior sales representatives’jobs.Compared with the traditional visit strategy,the new visit strategy of deep learning has achieved good results and increases the relative success rate of visit by about 120%.