Timeline for How to evaluate chi squared result?
Current License: CC BY-SA 3.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 30, 2018 at 3:49 | comment | added | Meir Maor | Calling something deterministic random is problematic. We do not currently know if pi is normal. Meaning the frequency of digitsand digit sequences is uniform. | |
| Mar 30, 2018 at 2:57 | comment | added | Paul Uszak | As to probability, you can't mathematically have p(isRandom)=100%. There is no statistical test in the world that can prove randomness. We can only disprove it. We can't prove that $ \pi $ digits are random, we just suspect it very very much. | |
| Mar 30, 2018 at 2:42 | comment | added | Paul Uszak | @valerio_new No absolutely not - you should never have a chi=0. That's not optimal. If chi=0, then you have an exactly uniform distribution of bytes in your sample. So there would be exactly $ \frac{1}{256} $ number of each value of byte. The chances of that happening with a large sample are zero. Most likely your generator is simply broken in that case. Read over the purpose of the chi test and how you applied it to random fluctuations of data results. And try to think of the shape my piccy would have to be so that chi=0 in both red and green squares. | |
| Mar 30, 2018 at 2:36 | history | edited | Paul Uszak | CC BY-SA 3.0 |
Link fixed.
|
| Mar 29, 2018 at 23:26 | comment | added | valerio_new |
Okay, but let's say, just for argument's sake, that my values tend to optimal, meaning that $(o_i - e_i)^2$ tends to zero. In this hypotesis, why my chi squared wouldn't be zero? Or would it? And why i should not be okay with that (assuming an insanely big sample)? > Your 254.3 value simple means that, meh, it's exactly 50/50 that your particular sequence is random. And why should i be okay with a 50/50 probabiliy of random? Shouldn't i look for the highest? Thank you! (the Kolmogorov–Smirnov link seems to be broken)
|
|
| Mar 29, 2018 at 15:02 | history | edited | Paul Uszak | CC BY-SA 3.0 |
Stuff about Binomials
|
| Mar 29, 2018 at 14:23 | history | edited | Paul Uszak | CC BY-SA 3.0 |
Pi p value
|
| Mar 29, 2018 at 12:07 | history | edited | Paul Uszak | CC BY-SA 3.0 |
added 2 characters in body
|
| Mar 29, 2018 at 12:02 | history | answered | Paul Uszak | CC BY-SA 3.0 |