Professional Documents
Culture Documents
Now by using the Minitab option that allows the user to specify the number of bins, the histogram has
been constructed as in figure 1.
To identify the distribution, well go to Stat > Quality Tools > Individual Distribution Identification in
Minitab. This handy tool will enable us to easily compare how well the data fit 16 different distributions.
In addition to using the probability plots to compare between the different types of distributions:
Its generally valid to compare p-values between distributions and go with the highest. A low p-value
(e.g., < 0.05) indicates that the data dont follow that distribution [1]. Furthermore if the AD value is
small, that another indication of a better fit, and for 3-parameter distributions only a low value of LRT P
indicates that adding the third parameter is a significant improvement over the 2-Parameter version.
As we can see from the Goodness of Fit Test table above:
The very first line shows our data are definitely not normally distributed, because the p-value for Normal
is less than 0.05.
Which is the case of the other distributions except the 3-Parameter Weibull, which has the highest p-
value (0.109), with lower AD and, the LRT P is significant (0.000), which means that the third parameter
significantly improves the fit.
Thats why we can choose the 3-Parameter Weibull distribution as the best fit for our data.
Fig. 3 The Probability Plots of (a) Normal distribution (b) Weibull distribution (c) Exponential distribution
(d) Lognormal distribution (e) Gamma distribution (f) 3-Parameters Weibull distribution
Its very clear that the 3-Parameter Weibull distribution in fig. 3 (f) is the best fit for our data, while the
rest of the distributions dont fit the data.
Ans: As we have 80 observations, we will choose the number of bins approximately equal to the square
root of the number of observations:
Now by using the Minitab option that allows the user to specify the number of bins, the histogram has
been constructed as in figure 5.
Comment: It can be clealy seen that the histogram has 8 bins. While we created it by selecting 9 bins
manually. However, if 9 bins are not specified, Minitab generates 10-bins histogram as in figure 6. As we
have mentioned that this formula is an approximation, and therefore either 8 or 10 bins should be enough
for assessing the distribution of the data.
c. Convert the stem-and-leaf plot in part (a) into an ordered stem-and-leaf plot. Use this
graph to assist in locating the median and the upper and lower quartiles of the viscosity data.
Ans: By using Minitab: MTB > Graph > Stem-and-Leaf, because Minitab is automatically creating an
order stem and leaf
For Q1:
(0.25)(80) + 0.5 = 20.5 (halfway between the twentieth and twenty first observation) which is:
(14.3 + 14.3)
= 14.3
2
For Q3:
(0.75)(80) + 0.5 = 60.5 (halfway between the sixtieth and sixty first observation) which is:
(15.6 + 15.5)
= 15.5
2
For Median:
(0.5)(80) + 0.5 = 40.5 (halfway between the fortieth and forty-first observation) which is:
(14.9 + 14.9)
= 14.9
2
From figure 7, the normal probability plot, we can clearly see that the data points do not fall along the
straight line, which means that the normal distribution does not reasonably describe process yield.
(a)
(b)
(c)
Fig. 8 Probability Plot of Viscosity Data (a) Normal (b) Lognormal (c) Weibull
Ans: From figure 8 we can see that both the normal and lognormal distributions suitable to be
reasonable models for the data; where the plot points fall along the straight line, without bends or
curves. While, the plot points on the Weibull probability plot are not straightparticularly in the tails
which means, it is not a reasonable model.
Also we can calculate it, since we have hypergeometric distribution with N = 25 and n = 5, without
replacement: then:
--------------------------------------
b. Calculate the desired probability in (a) using the binomial approximation. Is this
approximation satisfactory? Why or why not?
Ans: The binomial approximation to the hypergeometric:
To consider as a good approximation, the approximation has to satisfy the following condition:
----------------------------------------
c. Suppose the lot size was N = 150. Would the binomial approximation be satisfactory in this
case?
Ans: If N = 150, then, n/N = 5/150 = 0.033 0.1, As a result the binomial approximation would be a
satisfactory approximation to the hypergeometric in this case.
----------------------------------------
Q6:
Ans: Poisson distribution with = 0.01 errors/bill and x=1 :
e x e0.01 (0.01)1
p(x) = x!
Pr{ = 1} = (1) = 1!
= 0.0099
Then the probability that a customers bill selected at random will contain one error, is 0.99%
1 (0.0228) = 2
32
= -2 = 8 + 32 = 40
4