Timeline for True probability vs estimated probability
Current License: CC BY-SA 4.0
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 16, 2018 at 7:40 | comment | added | Fabian Werner | To make a very long and mathematically not always precise story short: Lets say $\Theta$ is your model parameter and $X$ is the data (probably involving input features, true answers and so on) then frequentists maximize $p(x|\theta)$. Bayesians also put the assumption of the form of $p(\theta)$ into the whole model and then optimize $p(\theta|x) = p(x|\theta)p(\theta)/p(x)$. Since you have not yet stated something about $p(\theta)$ I guess that you still follow the frequentists approach so far... | |
| Aug 16, 2018 at 7:16 | comment | added | user_anon | I have a related question: when assuming something (eg equally likely outcomes), we "are" bayesians or frequentists? I thought the latter, but our assumption is subjective I guess so we might be bayesians... | |
| Aug 14, 2018 at 11:18 | comment | added | Fabian Werner | Yes, you are right. In order to fit a distribution you need 'much' empirical evidence (mostly in form of data). | |
| Aug 14, 2018 at 10:45 | vote | accept | user_anon | ||
| Aug 14, 2018 at 10:32 | comment | added | user_anon | As a conclusion, the probability cannot be known, but can be assumed to be some number. My two examples were not examples of estimating probabilities (because there was NO data: I incorrectly said that the marbles were data...), but of making someone to think of a specific assumption: equally likely outcomes assumption. Right? | |
| Aug 14, 2018 at 9:44 | history | answered | Fabian Werner | CC BY-SA 4.0 |