| Autor: |
Kubelka, Vít |
| Přispěvatelé: |
Anděl, Jiří, Antoch, Jaromír |
| Jazyk: |
čeština |
| Rok vydání: |
2014 |
| Předmět: |
|
| Popis: |
The chi-squared goodness-of-fit test depends on the choice of classes. It is the purpose of this thesis to investigate differences between p-values depending on the choice of classes in the case of Poisson distribution. Samples from Poisson distribution of various sizes are simulated. For all choices of classes which satisfy condition of minimal theoretical frequency p-value is calculated. The simulations are made for various minimal theoretical frequencies. The unknown parameter is estimated both by modified method of minimum chi-squared and by sample mean. The simulations show that the condition of minimal theoretical frequency of classes is not sufficient and the tests are unreliable. Therefore the rule of almost equal theoretical frequencies in all classes is added and in the further simulation a substantial improvement of the test is shown when this condition is satisfied. Powered by TCPDF (www.tcpdf.org) |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
