TY - JOUR
T1 - Analyzing survival data with highly negatively skewed distribution
T2 - The Gompertz-sinh family
AU - Cooray, Kahadawala
AU - Ananda, Malwane M.A.
PY - 2010/1
Y1 - 2010/1
N2 - In this article, we explore a newtwo-parameter family of distribution, which is derived by suitably replacing the exponential term in the Gompertz distribution with a hyperbolic sine term. The resulting new family of distribution is referred to as the Gompertz-sinh distribution, and it possesses a thicker and longer lower tail than the Gompertz family, which is often used to model highly negatively skewed data. Moreover, we introduce a useful generalization of this model by adding a second shape parameter to accommodate a variety of density shapes as well as nondecreasing hazard shapes. The flexibility and better fitness of the new family, as well as its generalization, is demonstrated by providing well-known examples that involve complete, group, and censored data.
AB - In this article, we explore a newtwo-parameter family of distribution, which is derived by suitably replacing the exponential term in the Gompertz distribution with a hyperbolic sine term. The resulting new family of distribution is referred to as the Gompertz-sinh distribution, and it possesses a thicker and longer lower tail than the Gompertz family, which is often used to model highly negatively skewed data. Moreover, we introduce a useful generalization of this model by adding a second shape parameter to accommodate a variety of density shapes as well as nondecreasing hazard shapes. The flexibility and better fitness of the new family, as well as its generalization, is demonstrated by providing well-known examples that involve complete, group, and censored data.
KW - Gompertz distribution
KW - Goodness-of-fit
KW - Maximum likelihood
UR - http://www.scopus.com/inward/record.url?scp=72149094044&partnerID=8YFLogxK
U2 - 10.1080/02664760802663072
DO - 10.1080/02664760802663072
M3 - Article
AN - SCOPUS:72149094044
VL - 37
SP - 1
EP - 11
JO - Journal of Applied Statistics
JF - Journal of Applied Statistics
SN - 0266-4763
IS - 1
ER -