From: Evaluation of threshold selection methods for adaptive kernel density estimation in disease mapping
Desirable characteristics | Density curve characteristics | Age groups | ||||
---|---|---|---|---|---|---|
35–44 | 45–54 | 55–64 | 65+ | 35+ | ||
Most ↓ Least | Density curve is smoother, and fluctuations in the tail ceases to exist | h ns | h ns | h ns | h ns | 500 |
Median | Median | Median | Median | h ns | ||
500 | 500 | 500 | 500 | 1000 | ||
Density curve closely matches to the actual gamma distribution and contains fluctuations at the tail | 100 | 100 | 100 | 100 | h pi h scv | |
The highest density estimates of density curve is greater than that of the actual gamma distribution, and the density curve contains high fluctuations at the tail | 50 | 50 | 50 | 50 | 50 | |
h pi | h pi | h pi | h pi | 100 | ||
h scv | h scv | h scv | h scv | |||
Density curve is smoother and difficult to distinguish between the mode and tail | 1000 | 1000 | 1000 | 1000 | Median | |
Density curve is flat and cannot distinguish between the mode and tail | 5000 | 5000 | 5000 | 5000 | 5000 | |
10,000 | 10,000 | 10,000 | 10,000 | 10,000 |