Timeline for approximation of a known distribution by another distribution
Current License: CC BY-SA 3.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 26, 2017 at 0:35 | vote | accept | quibble | ||
| May 25, 2017 at 14:08 | comment | added | Tim | The basic optimization is to use EM algorithm rather then native black-box optimization. EM is most commonly used algorithm for mixtures and is very good for such case. | |
| May 25, 2017 at 8:33 | answer | added | Glen_b | timeline score: 4 | |
| May 25, 2017 at 8:26 | comment | added | quibble | In the sum of squared example above, I know it does not work because the same amount of error probably should be penalized more heavily if it occurs at a value with a high probability, but I don't know if there are appropriate weights for the penalization. | |
| May 25, 2017 at 8:06 | history | edited | quibble | CC BY-SA 3.0 |
clarification
|
| May 25, 2017 at 8:00 | comment | added | quibble | (In the example) Better way to estimate the parameters of the mixture distributions. I am not trying to find a distribution that better approximates the negative binomial distribution than the mixture poisson distribution. | |
| May 25, 2017 at 7:51 | comment | added | Tim | Better way to do what exactly? Estimate parameters of mixture? Better alternative for mixture? | |
| May 25, 2017 at 7:26 | history | asked | quibble | CC BY-SA 3.0 |