兄弟，你的泊松分布的验证解决了没。求助啊。。。。

Usage Note 22427: Can I test for Uniform distribution?
Details        About        Rate It       

Yes. The Uniform distribution is a special case of the Beta distribution with α = β = 1. You can test that the data are from a uniform distribution in PROC UNIVARIATE or in SAS/QC PROC CAPABILITY. In any of the statements that accept distribution options (such as the HISTOGRAM statement), use the options A=1 and B=1 with the BETA distribution option to test for uniformity. Use the THETA= and SIGMA= options to establish the lower and upper limits of support. Set THETA= to the lower limit. Set SIGMA= to the range between the upper and lower limits. For instance, to test that the input data are distributed U(10,25), specify the option:   BETA(A=1 B=1 THETA=10 SIGMA=15).
------------------------------------------------------------------------------------------------------------------------------
For poissson
Re: How to test Poisson Distribution?

Paul R Swank
Tue, 28 May 2002 11:47:07 -0700

Why can't you run a poisson regression model with no predictors and look at
the fit of the model. I generated 1000 Poisson x's and then fit to a poisson
using genmod in sas. The Pearson chi-square below is not significant
indicating fit. the mean is exp(intercept) = exp(-0.0253) = .975 which is
close to the mean of 1 that I specified.

                                    The GENMOD Procedure

                                      Model Information

                               Data Set              WORK.TEMP1
                               Distribution             Poisson
                               Link Function                Log
                               Dependent Variable             y
                               Observations Used           1000


                            Criteria For Assessing Goodness Of Fit

                 Criterion                 DF           Value
Value/DF

                 Deviance                 999       1143.3883
1.1445
                 Scaled Deviance          999       1143.3883
1.1445
                 Pearson Chi-Square       999        993.2051
0.9942
                 Scaled Pearson X2        999        993.2051
0.9942
                 Log Likelihood                     -999.6849


          Algorithm converged.


                               Analysis Of Parameter Estimates

                                  Standard     Wald 95% Confidence
Chi-
   Parameter    DF    Estimate       Error           Limits
Square    Pr > ChiSq

   Intercept     1     -0.0253      0.0320     -0.0881      0.0375
0.62        0.4292
   Scale         0      1.0000      0.0000      1.0000      1.0000
NOTE: The scale parameter was held fixed..
Paul R. Swank, Ph.D.
Professor, Developmental Pediatrics
Medical School
UT Health Science Center at Houston



-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On
Behalf Of Chia C Chong
Sent: Sunday, May 26, 2002 5:38 PM
To: [EMAIL PROTECTED]
Subject: How to test Poisson Distribution?


I have a random variable,X with 1000 samples (all discrete values). I want
to test whether X can sastify a Poisson process or not. How should I test
it?

THanks.
CCC


.
.
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .

--------------------------------------------

For bionomial distribution
http://www.ats.ucla.edu/stat/sas/whatstat/whatstat.htm
Binomial test

    A one sample binomial test allows us to test whether the proportion of successes on a two-level categorical dependent variable significantly differs from a hypothesized value.  For example, using the hsb2 data file, say we wish to test whether the proportion of females (female) differs significantly from 50%, i.e., from .5.  We will use the exact statement to produce the exact p-values.

    proc freq data = "c:\mydata\hsb2";
      tables female / binomial(p=.5);
      exact binomial;
    run;

    The FREQ Procedure

                                       Cumulative    Cumulative
    female    Frequency     Percent     Frequency      Percent
    -----------------------------------------------------------
         0          91       45.50            91        45.50
         1         109       54.50           200       100.00

    Binomial Proportion for female = 0
    -----------------------------------
    Proportion (P)               0.4550
    ASE                          0.0352
    95% Lower Conf Limit         0.3860
    95% Upper Conf Limit         0.5240

    Exact Conf Limits
    95% Lower Conf Limit         0.3846
    95% Upper Conf Limit         0.5267

       Test of H0: Proportion = 0.5

    ASE under H0                 0.0354
    Z                           -1.2728
    One-sided Pr <  Z            0.1015
    Two-sided Pr > |Z|           0.2031

    Exact Test
    One-sided Pr <=  P           0.1146
    Two-sided = 2 * One-sided    0.2292

    Sample Size = 200

    The results indicate that there is no statistically significant difference (p = .2292).  In other words, the proportion of females in this sample does not significantly differ from the hypothesized value of 50%.

====================================================

There are other tests, pleae refer to http://www.ats.ucla.edu/stat/sas/whatstat/whatstat.htm

thanks for sharing

谢谢啦！