Sei sulla pagina 1di 8

MENG 3402: Quality and process control

Lab report #2
Dr. Hatem El Ayat

Name: Amr M. Zahran

ID: 900120577

Table of Contents
Procedure................................................................................. 2
Conclusion................................................................................ 4
References...............................................................................5
Appendix.................................................................................. 6

List of Figures
Figure 1: Normal Probability Plot..............................................................................................3

Introduction
Normal probability plot is a graphical technique used for assessing whether or not a
data set is approximately normally distributed. The data are plotted against a theoretical

normal distribution in such a way that the points should form an approximate straight line.
Departures or variations from this straight line indicate departures from normality. The
normal probability plot is a special case of probability plots that is usually studied due to its
importance in several applications. A sample normal probability plot is shown below in figure
(1).
The Anderson-Darling test is a statistical test of whether a given sample data is drawn
or extracted form a given probability distribution. In its basic form the Anderson-Darling test
assumes that there are no parameters to be estimated in the distribution being tested, in such
case the test and its set of critical values is distribution free. It is a modification of the
Kolmogorov-Smirnov (K-S) test and gives more weight to the tails than does the K-S test.
This has the advantage of allowing a more sensitive test and the disadvantage that critical
values must be calculated for each distribution.

Figure 1: Normal Probability Plot.

Procedure

The thickness (height) of a sample of 50 metallic parts were measured using a


digital micrometer and the results were tabulated in an excel sheet. [done

through the first lab].


The digital micrometer was re-calibrated after each 10 measurements to make

sure that the readings were as accurate as possible.


The excel sheet is then imported into Excel.
The command attach(header) was then used to attach the imported data to

the active R session.


(nortest) package is or library is then loaded in R.
Anderson-Darling normality test is performed on the imported data.
Graphical normal probability plot (qq-plot) is also performed on the data and

the results are visually inspected.


The last test performed on the data is the Pearson test [Chiq-square] goodness

of fit test for normality.


Plots are exported into PDF format.

Conclusion
After performing the Anderson-Darling normality test on R the resulting p-value for
this set of data was 0.0141 which is clearly less than 0.05 indicating that the AndersonDarling test rejects the normality assumption. The QQ-plot shown in figure (1) was then
plotted in R. The data is clearly not perfectly fitting the normality distribution line which is
also an indication for rejecting the normality assumption for this data. The last test performed
was the Pearson Chi-square normality test which yielded a P-value = 0.0214 which is also
less than 0.05 so the Pearson Chi-square test also confirms that the data at hand does not
follow a normal distribution.

References
1] Pearson E.S., Hartley, H.O. (Editors) (1972) Biometrika Tables for Statisticians,
Volume II. CUP. ISBN 0-521-06937-8.
2] Shapiro, S.S. (1980) How to test normality and other distributional assumptions. In:
The ASQC basic references in quality control: statistical techniques 3, pp. 178.
3] Shorack, G.R., Wellner, J.A. (1986) Empirical Processes with Applications to
Statistics, Wiley. ISBN 0-471-86725-X.
4] Stephens, M.A. (1979) Test of fit for the logistic distribution based on the
empirical distribution function, Biometrika, 66(3), 5915.

Appendix
> Metallic.parts <- read.table("~/Desktop/Metallic parts.csv", quote="\"", comment.char="")
> View(Metallic.parts)
> attach(Metallic.parts)
> ad.test(V1)
Anderson-Darling normality test
data: V1
A = 0.96095, p-value = 0.0141
> qqnorm(V1)
> qqline(V1,col="red")

> pearson.test(V1)
Pearson chi-square normality test
data: V1

P = 16.4, p-value = 0.0217


Metallic Parts Table (dimensions measured in the first lab):
10.538
9.73
10.38
9.363
9.92
10.789
9.952
9.751
10.291
10.427
10.431
9.962
9.924
10.135
10.015
10.115
9.902
9.932
10.205
9.98
10.31
10.014
9.351
9.903
9.873

10.522
10.017
9.91
9.803
10.066
10.11
9.81
9.749
9.924
9.785
10.553
9.909
10.257
9.919
10.612
10.051
10.134
10.047
9.981
10.487
9.901
10.531
9.751
10.096
10.131

The order of the data is by moving downwards through the first column then moving
downwards through the second column.

Potrebbero piacerti anche