论文部分内容阅读
在营销管理中,消费者的选择行为预测,对适当的消费者细分和找到目标消费者具有很大的价值。通过引进转移学习的思想,然后基于该思想使用较小的数据样本,对消费者选择行为进行预测,并构建一个新的选择预测模型——转移装袋模型,从不同地区的相似消费者中提取信息,以提高选择模型的预测能力。用该模型进行实证分析,即用户选择何种电话卡的行为作分析预测,并与其他预测模型作了对比。结果表明,新模型的选择预测能力明显高于其他模型。
In marketing management, consumer choice behavior prediction, the appropriate consumer segmentation and find the target consumer has great value. By introducing the idea of transfer learning and then using the smaller data samples based on the idea, this paper predicts consumer choice behavior and constructs a new choice forecasting model - transfer bagging model, which extracts from similar consumers in different regions Information to improve the predictive power of the selection model. This model is used to conduct empirical analysis, that is, what kind of calling card users choose for their behavior analysis and prediction, and compared with other forecasting models. The results show that the ability of the new model to forecast is obviously higher than that of other models.