Summary: | 碩士 === 國立嘉義大學 === 應用數學系研究所 === 101 === The change point problems occur frequently in economic, medical, and industrial. For example, the arteriosclerosis will surge when over 55 years old or when a stock over a transaction prices, the volume will be decrease and this particular value was named the change point. So if we can accurately estimate the change point, it is
conducive to prevent this disease or to analysis on stock.
There are many methods, such as MLE and LSE, to estimate the parameters in linear regression model, however, MLE is useless in piecewise regression since it is everywhere continues but not be differentiable in the change point. This research discusses the piecewise regression model with multiple change points, not only uses a smooth transition function into segmented regression models, but also uses the LSE method to deal with discontinuous piecewise regression model. This study found that
a smooth transition function is more appropriate than the LSE method.
|