| ABSTRACT | 第3页 |
| 1 INTRODUCTION | 第10-15页 |
| 1.1 Research Background | 第10-11页 |
| 1.2 The scope of this research | 第11-12页 |
| 1.3 Motivation for this research | 第12-13页 |
| 1.4 The structure of this thesis | 第13-15页 |
| 2 LITERATURE REVIEW | 第15-28页 |
| 2.1 Multidimensional observation | 第15页 |
| 2.2 Brief introduction to multivariate data and analysis | 第15-17页 |
| 2.3 Pre-processing for Multivariate Data | 第17-19页 |
| 2.4 Visualiazation of Multivariate Data | 第19-20页 |
| 2.5 Multiple vs. Multivariate Time Series Data Analysis | 第20-21页 |
| 2.5.1 Understanding the nature of the data | 第21页 |
| 2.6 Multivariate Time Series: Our research point of view | 第21-22页 |
| 2.7 Stationarity and auto-covariance functions | 第22-23页 |
| 2.8 Cross-correlation function | 第23-24页 |
| 2.9 Multivariate white noise | 第24-25页 |
| 2.10 Examining anomalies or outliers | 第25-28页 |
| 2.10.1 Anomaly Detection | 第25-26页 |
| 2.10.2 Applications of Anomaly Detection | 第26-28页 |
| 3 MODELING | 第28-54页 |
| 3.1 An Ensemble Multivariate Model for Resource Performance Predictionin the Cloud | 第28-36页 |
| 3.1.1 Prediction theory and techinques | 第29页 |
| 3.1.2 Prediction theory | 第29-30页 |
| 3.1.3 Prediction techniques | 第30页 |
| 3.1.4 Stable Vector Autoregressive Model | 第30-32页 |
| 3.1.5 Dynamic Linear Models | 第32-33页 |
| 3.1.6 Ensemble learning approach | 第33-36页 |
| 3.2 Investigating Multivariate Resourses Outliers Detection | 第36-40页 |
| 3.2.1 Data distribution | 第36-39页 |
| 3.2.2 Correlation among variables | 第39-40页 |
| 3.3 Cloud Computing, Ensemble Multivariate Resource Performance basedon Machine Learning Techniques | 第40-54页 |
| 3.3.1 Support Vector Machine (SVM) | 第40-43页 |
| 3.3.2 K-Nearest Neighbors | 第43-47页 |
| 3.3.3 Random Forest | 第47-49页 |
| 3.3.4 Neural Network | 第49-50页 |
| 3.3.5 Improving prediction performance | 第50-51页 |
| 3.3.6 Optimization with Hill Climbing Approach | 第51-52页 |
| 3.3.7 Introduced Model | 第52-54页 |
| 4 EXPERIMENTS | 第54-62页 |
| 4.1 Data selection | 第54页 |
| 4.2 Root Mean Square Error | 第54-55页 |
| 4.3 Mean Absolute Error | 第55页 |
| 4.4 Granger Causality | 第55-57页 |
| 4.4.1 Personal account by Clive Granger(This account is meant not to be changed/edited from its original) | 第55页 |
| 4.4.2 Limitations and extensions: Linearity | 第55-56页 |
| 4.4.3 Limitations and extensions: Stationarity | 第56页 |
| 4.4.4 Limitations and extensions: Dependence on observed variables | 第56页 |
| 4.4.5 Granger causality analysis for a part of this research | 第56-57页 |
| 4.5 Ensemble Learning | 第57页 |
| 4.5.1 Ensemble Learning analysis for this part of research | 第57页 |
| 4.6 Robust methods for Multivariate outlier detection | 第57-60页 |
| 4.6.1 Normal quantile plot | 第57-58页 |
| 4.6.2 Mahalanobis distance | 第58-59页 |
| 4.6.3 Robust Location and Scatter Estimators for Multivariate DataAnalysis | 第59页 |
| 4.6.4 The computation of the distances based on principal components | 第59-60页 |
| 4.6.5 Univariate Presentation of Multivariate Outliers | 第60页 |
| 4.7 Hill Climbing and Ensemble Learning Processes | 第60-62页 |
| 5 RESULTS AND DISCUSSION | 第62-80页 |
| 5.1 Evaluation of an Ensemble for Multivariate Model for Resource Perfor-mance Prediction in the Cloud | 第62-67页 |
| 5.1.1 Root Mean Square Error Evaluation | 第62页 |
| 5.1.2 Granger causality evaluation | 第62-67页 |
| 5.1.3 Ensemble Learning Evaluation | 第67页 |
| 5.2 Evaluation of Outliers Detection | 第67-73页 |
| 5.2.1 Adjusted Quantile plots | 第67-69页 |
| 5.2.2 Mahalanobis distance evaluation | 第69-70页 |
| 5.2.3 Robust location and scatter estimation via MCD Evaluation | 第70-71页 |
| 5.2.4 Robust principal components | 第71-72页 |
| 5.2.5 One-dimensional scatter plots | 第72-73页 |
| 5.3 Evaluation of Ensemble for Multivariate Resource Performance basedon Machine Learning Techniques | 第73-80页 |
| 5.3.1 RMSE and MAE evaluation | 第73-74页 |
| 5.3.2 Hill climbing and ensemble learning evaluation | 第74-80页 |
| 6 CONCLUSION | 第80-82页 |
| 7 REFERENCES | 第82-89页 |
| A APPENDIX | 第89-97页 |
| A.1 Data Transformation | 第89-92页 |
| A.1.1 Transformation of variables | 第89-90页 |
| A.1.2 Dividing Seasonal Standard Deviations | 第90页 |
| A.1.3 Subtracting Seasonal Means | 第90页 |
| A.1.4 Differencing | 第90-91页 |
| A.1.5 Regression on Trends and Cycles | 第91页 |
| A.1.6 Moving Average | 第91-92页 |
| A.2 Transformations for multivariate data | 第92-93页 |
| A.3 Multivariate vs Univariate Analysis Results | 第93-97页 |
| ACKNOWLEDGEMENT | 第97-98页 |
| PUBLISHED PAPERS | 第98-100页 |
