一种基于K-means的神经网络数据集回归预测算法OA

The standard statistical model cannot balance the information value across various dimensions which has an impact on the predictive power of data sets in the regression analysis and prediction model of high-dimensional data.This paper suggests a K-means regression prediction model based neural network data set.Firstly,at the feature level,the multi-layer RNN neural network extracts data features from several dimensions and trains the response.Secondly,the data is classified using the K-means classifier model at the algorithm level,which integrates the feature response of a Recurrent Neural Network(RNN)neural network,and a combined prediction model is built for the data set of the output response in order to increase the predictability of the algorithm.Finally,in such measure of regression analysis of time series data with multi-dimensional features,the experimental results of the UCI regression analysis data set compared to the Gated Recurrent Unit(GRU)algorithm demonstrate that this approach further enhances the definition of model prediction by 14.45 percent.