论文部分内容阅读
针对信息客户流失缺少有效的数据挖掘预测手段问题。提出了应用主成分分析的BP神经网络信息流失预测模型。结合5折交叉验证,模型对来自3个地市的营销返回样本,在训练分类时间和预测分类精度上与未经主成分分析降维的BP神经网络方法进行了比较分析。实验结果表明模型获取了较高的平均预测分类精度(77.46%)和较少的训练分类时间(2.18min),有效地降低了属性维度并改善了预测能力。
Lack of effective and effective data mining forecast means for information loss of customers. Proposed a principal component analysis BP neural network prediction model of information loss. Combined with 5-fold cross-validation, the model returns samples from 3 cities, and compares them with the BP neural network method without dimensionality reduction in training classification time and prediction classification accuracy. The experimental results show that the model achieves a high average prediction accuracy (77.46%) and a small training classification time (2.18min), effectively reducing the attribute dimension and improving the prediction ability.