Curves Clustering Based on Quantile Regression

摘要

The motivation of time series classification is to find similar volatility structures, reduce the workload and forecast, hence the outcome of classification would directly impact on the quality of models and the accuracy of forecasts. For this purpose, the paper proposed a new method of time series classification QRP Clustering. QRP Clustering can avoid some limitations brought by several classification methods, fully test the operation of time series waiting to be classified, improve the effectiveness of classification and provide strong support for forecasts.