Research On Big Data Multiple Regression Prediction Method Based On Hadoop

With the advent of the era of "+" mutual network,the application of network computers has generally become the focus of attention.Internet technology has already penetrated into every fields of human society.The development of cloud computing and all kinds of new social networking platform makes the amount of data and data types in the human society expand.The storage,management and analysis of massive data have brought people great challenges.Cloud computing will be able to use a lot of cheap computer clusters to replace the expensive servers,which to a great extent reduces the computational expenses.The appearance of the Hadoop provides an excellent platform for people to solve huge amounts of data.Compared with the traditional data processing platform,not only is it able to store,dispose,analyze large amounts of data,but also obtains a better effect on the processing speed and computing performance.Based on the advantages of disposing large amounts of data by Hadoop distributed platform,this paper comes up with Hadoop multiple regression prediction method and applies it to the prediction of mooring fender pressure so that it can improve the computational efficiency of predicting the mooring fender pressure and enhance its applicability and commonality.As to today' s private computing of huge amounts of data linear regression for too long,this paper has designed and realized a parallel multiple linear regression algorithm of graphs.On Hadoop platform,with the HBase database realizing the large file storage and fault toleran ce,with the MapReduce distributed algorithm pattern processing the data in the HBase database,combined with the multiple regression algorithm to forecast the mooring fender pres sure data,with the gradient descent algorithm optimizing the regression coefficient,the true solution will be obtained.Finally,through the Hadoop computing platform and database we'll get the results after calculating.Through the experiment,in the application of Hodoop platform and HBase database to the prediction of mooring fender pressure,the efficiency and accuracy of predicting the ship's mooring fender pressure have been greatly improved by multiple regression forecast method.