The Lubricating Oil Production Prediction Model Based On Robust Principal Component Regression

In recent years,with the rapid development of our country's automobile,engineeringmachinery,machine tool industry,electric power,steel and other industries,the improvement of the equipment technology,the demand for lubricating oil in our country is increasing.Our country has become the world's second largest lubricating oil market.However,domestic demand is not the same as domestic production,before the 1990 s,domestic lubricating oil production and sales are controlled by the country.With the development of energy system reform,lubricating oil has become one of the earliest release of petrochemical industry.Domestic lubricant brand's market share is higher and higher.But because of many international brands into China's lube oil market,domestic lubricant brand will be taken to the road of the intensification of competition.At the same time,with the continuous development of our country's economy,lubricating oil foreign markets are also expanding,quilt a number of lubricating oil have been exported to Asia and other countries.In order to effectively control the market situation and prevent the lubricating oil production being insufficient,it is particularly important to predict monthly lubricating oil production.The lubricating oil production figures which are found in the national bureau of statistics website show that some data gravely deviates from the other data.That is to say there are outliers in data of oil output.For the accuracy of the estimation,a robust estimation method is considered.At the same time,to predict the lubricating oil production accurately,it is very important to find out the factors which influence the production of lubricating oil.Because the information expressed by many factors overlap,that is to say,multillinearity exists,it is considered to use principal component analysis.So in this article I will use the robust principal component regression method for lubricating oil production prediction.So-called robust principal component regression is to combine principal component estimation,M estimation with multiple regression analysis to establish the model of lubricating oil production prediction.In this way,the situations that the dependent variable has outliers and multillinearity exists can be handled.