There are few structural break-point analysis methods like Chow, Quandt-Andrews, Zivot-Andrews, Additive & Innovative outlier methods. Other side, non-linear procedures have also been followed. Which is advantageous and in what aspects?
Thnx Balazs. I,m studying trends in major crops in India. Scatter shows few crops have 1 break point and few have more than 1. I wish to go for a common method of estimation. Trend is mostly linear and exponential in the study period. And one has quadratic.
Recursive entropic segmentation and the mean- field analysis of it may be useful.Taking an approach that is often used in DNA segmentation. Auto correlative properties as horizontal "DNA transfers". DNA sequences as structures within otherwise bewildering mosaic of DNA information. Might need an idea of the ideal sequence, template, to be able to gauge the current data set's entropy at given points.Works particularly well in economic time series that is complex and volatile and of a fine-grained nature. Or noisy data as we take a macroscopic view/integral view. You might want to call it a data science intensive approach to an otherwise straightforward econometric problem.
Article Extending the Recursive Jensen-Shannon Segmentation of Biolo...
Conference Paper Recursive Segmentation Procedure Based on the Akaike Informa...
Article The Japanese Economy in Crises: A Time Series Segmentation Study
- Badges
- Science topic
- Similar topics
- Psychology