Nonparametic [i.e. Nonparametric] estimation of density functions in hydrology | Neparametarske funkcije raspodele u hidrologiji
2011
Marković, Đ., Fakultet tehničkih nauka, Kosovska Mitrovica (Serbia) | Plavšić, J., Građevinski fakultet, Beograd (Serbia) | Stanić, M., Građevinski fakultet, Beograd (Serbia) | Sekulić, G., Građevinski fakultet, Podgorica (Montenegro)
As an alternative procedure for parametric methods in statistical analysis, non-parametric methods are emerging mainly because of their flexibility and ability to completely automate the estimation of probability density functions. One of the main features imposed by this method, in regard to parametric methods, is the absence of need to choose distribution based on samples or combination of two theorectical distribution functions for distribution tails. The paper presents advantages and disadvantages of kernel based probability distribution estimates on an example of fitting distributions of weekly flows for the purpose of simulating long hydrologic series. In this context, the non-parametric approach is particularly useful because of the possibilities to automate the whole process.
显示更多 [+] 显示较少 [-]