This paper deals with the estimation in nonparametrio regression model.Sincethe conditional mean is sensitive to the tail behavior of the conditional distributionof the model,instead conditional median is considered.F...This paper deals with the estimation in nonparametrio regression model.Sincethe conditional mean is sensitive to the tail behavior of the conditional distributionof the model,instead conditional median is considered.For estimation of theconditional median,the sequence of the nearest neighbor estimators is shown to beasymptotio normal and consistent.展开更多
We investigated the correlation of large fires（[300 ha） from 1992 to 2013 within the borders of the Antalya Regional Directorate of Forestry using the Keetch–Byram drought index（KBDI）. Daily KBDI values were calc...We investigated the correlation of large fires（[300 ha） from 1992 to 2013 within the borders of the Antalya Regional Directorate of Forestry using the Keetch–Byram drought index（KBDI）. Daily KBDI values were calculated for each year, and values for the period before the year 2000 differed significantly from those after2000. After 2000（large fires occurred in 2004, 2006, 2007,2008, 2010, and 2013）, when KBDI values increased, the KBDI, but not the number of fires, was inversely correlated with the natural log of the burned area（NLBA）. While there were both high and low KBDI values when the NLBA was small, only high KBDI values were associated with high NLBA values. Particularly for logarithmic values of 4 and higher, KBDI values increased in parallel with increases in NLBA values. On the basis of a Mann–Whitney U test done in addition to a Pearson correlation test, we found that when the burned areas were grouped according to small and large areas, the KBDI could be used to distinguish the two groups. Using a conditional probability analysis, we found that 4th, 5th and 6th class KBDI values may lead to large fires at the 60 % possibility.Similarly, the possibility of large fires greater than the median burned area in any given 6 years was found to be48 %. In addition, while the mean value of KBDI is 390.51 for the period from May to September for these 6 years, it is 359.93 for the other years. Consequently, the area burned also increased as the KBDI classes（Class 0： 0–99, Class 1：100–199, Class 2： 200–299, Class 3： 300–399, Class 4：400–499, Class 5： 500–599, Class 6： 600–699, and Class 7：700–800） increase.展开更多
文摘This paper deals with the estimation in nonparametrio regression model.Sincethe conditional mean is sensitive to the tail behavior of the conditional distributionof the model,instead conditional median is considered.For estimation of theconditional median,the sequence of the nearest neighbor estimators is shown to beasymptotio normal and consistent.
文摘We investigated the correlation of large fires（[300 ha） from 1992 to 2013 within the borders of the Antalya Regional Directorate of Forestry using the Keetch–Byram drought index（KBDI）. Daily KBDI values were calculated for each year, and values for the period before the year 2000 differed significantly from those after2000. After 2000（large fires occurred in 2004, 2006, 2007,2008, 2010, and 2013）, when KBDI values increased, the KBDI, but not the number of fires, was inversely correlated with the natural log of the burned area（NLBA）. While there were both high and low KBDI values when the NLBA was small, only high KBDI values were associated with high NLBA values. Particularly for logarithmic values of 4 and higher, KBDI values increased in parallel with increases in NLBA values. On the basis of a Mann–Whitney U test done in addition to a Pearson correlation test, we found that when the burned areas were grouped according to small and large areas, the KBDI could be used to distinguish the two groups. Using a conditional probability analysis, we found that 4th, 5th and 6th class KBDI values may lead to large fires at the 60 % possibility.Similarly, the possibility of large fires greater than the median burned area in any given 6 years was found to be48 %. In addition, while the mean value of KBDI is 390.51 for the period from May to September for these 6 years, it is 359.93 for the other years. Consequently, the area burned also increased as the KBDI classes（Class 0： 0–99, Class 1：100–199, Class 2： 200–299, Class 3： 300–399, Class 4：400–499, Class 5： 500–599, Class 6： 600–699, and Class 7：700–800） increase.
