4/14/99 STT 231
Directions For Minitab
1. Generating data from a specied ditribution.
To generate K samples of size N from distribution D:
(a) using command language:
random N c1-cK; (you can put them in any other K columns if you wish)
D…followed by the values of the parameters of D.
Examples: (i) random 20 c4-c8;
binomial 40 .77.
This generates 5 samples of size 20 each in columns c4-c8.
Each column contains a random sample of size 20 from a BIN(n,p)
with n=40 and p=.77
(ii) random 75 c11-c40;
normal 24 3.
This generates 30 samples of size 75 each in columns c11-c40.
Each column contains a random sample of size 75 from a N(,)
with = 24 and =3.
(b) using the Calc menu: Select ‘Random data’ and choose from among the various alternatives.
Sample From Columns - takes a random sample, with or without replacement of specified rows
Each command below generates random data from the distribution corresponding to the command name:
Chisquare Beta
Normal Poisson
F Cauchy
T Exponential
Uniform Gamma
Bernoulli Laplace
Binomial Logistic
Discrete Lognormal
Integer Weibull
When you click on one of these, you will be asked for the number of samples, where to put them, the sample sizes, and the parameter values of the model.
2. Getting the probabilitity f(k)=P[X=k] for a discrete model (the same method gets the value of the density for
continuous model )
(a) using commands: pdf k;
D…(params).
This will display (on screen) P[X=k] for the model specified by D and its parameters.
If you want to store the probability use: pdf k k1; The value will be stored as a constant in k1.
To see the value you have to issue a ‘print k1’ command.
You can obtain a table of values, stored or not, as follows:
Suppose that coloumn Cx contains a set of integers.
pdf Cx Cy;
D… (params).
This will produce a column, Cy, of P[X=k] for every integer k that is in Cx.
Example: c1 contains the integers 10-22.
MTB > pdf c1 c2;
SUBC> bino 30 .5.
MTB > print c1 c2
C1 10 11 12 13 14 15 16 17 18 19 20 21 22
C2 0.03 0.05 0.08 0.11 0.14 0.14 0.14 0.11 0.08 0.05 0.03 0.01 0.01
(b) using the Calcmenu: Select ‘Probability Distributions’ from the Calc menu. Fill in the requested information.
3. Getting the CDF : Proceed as in #2 above, replacing ‘pdf’ by ‘cdf’ (same Calc menu). This is for discrete or continuous models.
4. Getting percentiles: Proceed as in #2 above with these replacements (i) ‘pdf’ is replaced by ‘invcdf’;
(ii) the argument of the function is a probability, 0 < p < 1.
Example: invcdf .1
Normal 45 6.
This will produce the 10th percentile of a N(45,6) model.
5. Obtaining CI’s based on the t distribution: Assume you have data in columns Cx-Cy (any number of them).
To obtain P* percent CI’s (based on t) for each data set, use the command:
MTB > tint P* Cx-Cy
That’s all that’s needed. The screen will then display as many P* CI’s as you asked for. Remember that the t is used when the sd of the model is unknown. Minitab automatically estimates and uses the appropriate t-value based on the sample size. For problem 9 in HW4, P* = 80.