Modified Ridge Regression With Data Augmentation

The current society is experiencing the fourth revolution of science and technology,and data allows everything to be achieved.Data has become a new resource,such as water,electricity,land,air,forests,grasslands,minerals and so on.Statistical analysis as a method of processing data has become the focus in today's society that data is ubiquitous.And multi-linear model is one of the most widely used data analysis technology at present.In the time of 'big data',the complexity of the data and the rapid expansion of the dimension make it difficult for some traditional statistical analysis methods to adapt,such as OLS.The results of these original analytical methods gradually began to fail and can't be the reasonable explanation of the results.So there is a series of situations that evolve the new methods such as ridge regression,lasso,LARS and so on.Based on the basis,this paper introduces the method of ridge regression,including its background,basic theory,property and proof of the property.In addition,this paper also elaborates the selection method of the ridge regression's parameter k and lists some classical ridge regression's parameters k.Then the background and specific methods of Data Augmentation are summarized and expanded.Through two examples to fully demonstrate the Data Augmentation method has the applicability and ease of use as a supporting technology.On top of the above,in this paper,a modified ridge regression with Data Augmentation method is proposed.The results of the comparison with the data show that the modified method does improve the accuracy of prediction and the stability of prediction.