Application Of Multi-dimensional Emsemble Model With Big Data Prediction

With the coming of big data era,there is a growing demand for data mining and machine learning in various fields.On the one hand,in order to mine the low density and valuable information in the massive data,the classifier with high generalization and high accuracy is essential.Therefore,researchers continue to develop a variety of machine learning algorithms.Nowadays,the number of machine learning algorithms is increasing,and on this basis,more and more ensemble methods have been developed.On the other hand,with the increasing of data size,there are many dimensions,which also has certain prediction ability,to solve the same problem.To this problem,traditional machine learning algorithms and ensemble methods have been difficult to solve.In order to solve this problem,this thesis researches to propose a multi-dimensional ensemble model of big data prediction,which also has been applied to production and life.To find a better way to solve the above problem,this thesis considering that there are many factors influencing the predicting of football match result.Considering that each dimension can predict the result of football match independently,this thesis tries to select the Top Five European Football Leagues as the data sample to research.Through the study of the classification and prediction of the Top Five European Football Leagues,this thesis tries to find a general solution to the multi-dimensional ensemble model of classifier prediction.In thesis,data crawling technology will be used first,and then get predictable datasets for each dimension with data sorting,basic statistics of classification objectives,process missing and abnormal values,exploratory data analysis,feature engineering,etc.Then through the traditional machine learning algorithms and ensemble methods to classify and predict the result of each dimension.In addition,the thesis tries to solve the multi-dimensional prediction problem through the voting model.Finally,through exploring and innovating the improved stacking ensemble model,which based on the traditional machine learning model and ensemble model,achieves high accuracy and stability.Lastly,experimental and test results show the effectiveness of the improved stacking ensemble model.