Solutions to Exercises

Size: px
Start display at page:

Download "Solutions to Exercises"

Transcription

1 1 c Atkinson et al 2007, Optimum Experimental Designs, with SAS Solutions to Exercises 1. and 2. Certainly, the solutions to these questions will be different for every reader. Examples of the techniques that are required are given in the text of the book. Becoming an expert, of course, requires a lot of practice. 3. When the block effects are fixed, the condition for orthogonality defined in Chapter 15 requires the identities N 1 ij F ij T 1 Nij = N 1 F T 1 N, i = 1, 2,..., B; j = 1, 2,..., b i, (1) to be satisfied. In this case B = 2, b 1 = b 2 = 2, N = 18, N 11 = N 12 = 6, N 21 = N 22 = 3, and the columns of F correspond to the values of the regressors: the intercept, x and, when the model is second order polynomial, x 2, at each of the observations of the corresponding blocks, i.e F 11 = F 12 = and F 21 = F 22 = The average level of the regressors in an orthogonally blocked design is the same at each level of each blocking variable: zero for x and two thirds for x 2. Therefore the parameters corresponding to x and x 2 (if the model is of second order) will be estimated independently from the block effects when this design is used. It is easy to verify empirically the orthogonality of the design when the blocks are random. For example, the following program can be used. The blocking variables are z 1 and z 2. The sums of squares SS I and SS III are identical. Note that a block interaction cannot be estimated with this design.. data mydata; input z1 z2 x y;

2 2 cards; ; proc glm; class z1 z2; model y = z1 z2 x x*x / e1; random z1 z2; 4. A possible way to analyse the data is to run the SAS program given below. data Elastomer; do Oil = 0 to 30 by 10; do Filler = A, B, C ; do Level = 0 to 60 by 12; input logv = log(viscosity); output; end; end; end; datalines;

3 ; proc glm data=elastomer; where (Filler= A ); model logv = Level Oil; proc glm data=elastomer; where (Filler= A ); model logv = Oil Level ; proc glm data=elastomer; where (Filler= C ); model logv = Level Oil; proc glm data=elastomer; where (Filler= C ); model logv = Oil Level ; As the sums of squares SS I and SS III are identical for Filler C, the design is orthogonal. However, this is not true for Filler A as in this case the lost observation has destroyed this property of the planned design. Whether any of the proposed experimental designs is suitable depends on whether any further changes are planned for the new experiment. Provided that there are no such changes, a 2-level factorial design would be inappropriate to use as it would not allow for estimating a second order polynomial model. The points of the composite design when the star points are equal to -1 or 1 form a useful subset of the points of the 3-level factorial design. Such a design would be a good choice, when compared to a complete 3-level or higher order factorial design because it requires a relatively small number of observations to be taken. This design does not satisfy criteria 5 and 14, but satisfies criteria 1-4, 7, 10 and 13 from the criteria listed in Chapter 6. To some extend this also applies to criteria 8, 9 and 12, though if any of these is of particular concern, it might be possible to construct a better experimental design. An algorithmic search for a design would be appropriate when the available experimental resources require a number of observations different from that of the points of the composite design. This would also be a useful approach if the experimental scenario has changed and for example blocking is required, the design region has been changed or a different model has to be estimated. 5. Suppose that a quadratic term for the ith explanatory variable is significant. Such a model cannot be fitted as the column corresponding to the quadratic term in the extended design matrix consists of ones, exactly as that for the intercept in the model. Therefore ˆβ 0 is an estimate of β 0 + β ii. A lack of fit test would not pick up this potentially serious problem.

4 4 6. For simplicity of notation we rename the factors A, B,..., H. In order to obtain a design with 16 observations, 4 of the factors have to confounded by the design with high order interactions. For example, we can choose: E = BCD; F = ACD; G = ABC; H = ABD. The design can then be constructed by first writing down a complete 2-level factorial design for A, B and C; the values of the remaining factors can then be obtained using the above identities. As each factor is equal to ±1, its squared value is always equal to 1. Hence, we obtain I = ABCG = ABDH = ABEF = ACDF = ACEH = ADEG = AF GH = BCDE = BCF H = BDF G = BEGH = CDGH = CEF G = DEF H The intercept in the model is confounded with fourth order interactions. Therefore the estimate of the intercept is a sum of the effects I + ABCG + ABDH + ABEF + ACDF + ACEH + ADEG + AF GH +BCDE + BCF H + BDF G + BEGH + CDGH + CEF G + DEF H. If fourth order interactions are not significant, the intercept will be unbiased. In order to obtain the alias structure the above identity can be multiplied by one or more factors. No more than 15 further identities can be obtained. Each of them corresponds to a parameter that can be included in the model. Assuming that interactions of order higher than 3 are not significant, in a similar way as above we obtain: A + BCG + BDH + BEF + CDF + CEH + DEG + F GH B + ACG + ADH + AEF + CDE + CF H + DF G + EGH C + ABG + ADF + AEH + BDE + BF H + DGH + EF G D + ABH + ACF + AEG + BCE + BF G + CGH + EF H E + ABF + ACH + ADG + BCD + BGH + CF G + DF H F + ABE + ACD + AGH + BCH + BDG + CEG + DEH G + ABC + ADE + AF H + BDF + BEH + CDH + CEF H + ABD + ACE + AF G + BCF + BEG + CDG + DEF AB + CG + DH + EF AC + BG + DF + EH AD + BH + CF + EG AE + BF + CH + DG AF + BE + CD + GH AG + BC + DE + F H

5 5 AH + BD + CE + F G We would not be able to estimate effects that appear on the same rows above. As AB, AC and BC are on different lines they are estimable. (b). This becomes possible when we relabel G = x 8 and H = x 7. No more interactions would be estimable as the design becomes saturated. (c). Clearly the problem becomes more complicated. We need to choose a block generator, say AB, as well as 4 design generators, similarly to (a) above: E = BCD; F = ACD; G = ABC; H = ABD. We will no longer be able to estimate the same model as in (a) and (b). Estimable effects are listed below in a similar fashion to (b): I + ABCG + ABDH + ABEF + ACDF + ACEH + ADEG + AF GH +BCDE + BCF H + BDF G + BEGH + CDGH + CEF G + DEF H Block = AB + CG + DH + EF A + BCG + BDH + BEF + CDF + CEH + DEG + F GH B + ACG + ADH + AEF + CDE + CF H + DF G + EGH C + ABG + ADF + AEH + BDE + BF H + DGH + EF G D + ABH + ACF + AEG + BCE + BF G + CGH + EF H E + ABF + ACH + ADG + BCD + BGH + CF G + DF H F + ABE + ACD + AGH + BCH + BDG + CEG + DEH G + ABC + ADE + AF H + BDF + BEH + CDH + CEF H + ABD + ACE + AF G + BCF + BEG + CDG + DEF AC + BG + DF + EH AD + BH + CF + EG AE + BF + CH + DG AF + BE + CD + GH AG + BC + DE + F H AH + BD + CE + F G (d). If the interactions between x 1, x 2 and x 3 have to be estimated, we need to label them A, C and D in order to use the same design. However, it is not possible to estimate in addition two factor interactions of x 8 with x 4, x 5, x 6 and x 7. If such interactions are deemed significant, using this design would be inappropriate. 7. If 18 observations can be collected, it would still be possible to use one of the 16-trial designs described in Exercise 6. If the factors are continuous, the extra 2 observations can be taken at the centre of the experimental region, thus providing an opportunity to test for lack of fit of the model. Alternatively, an experimental design

6 6 Table 1: Central composite design for 3 explanatory variables x 1 x 2 x with 18 observations can be constructed using one of algorithms described in Chapters 12 and 13. What would be best depends on the experiment in which the design will be used. 8. Table 1 lists the points of a central composite design for 3 explanatory variables in standard order. If the observations are taken sequentially, their order should be randomized. This design would allow a second (or lower) order polynomial model to be estimated. 9. If a polynomial model is estimable when a complete or fractional two-level factorial design ξ is used, its information matrix is diagonal so that Hence the design is orthogonal. Also, F T F = NI p. d(x, ξ) = f T (x)f(x) = p fi 2 (x). This expression is maximised and equal to p when the elements of x are -1 or 1. When the absolute value of any element of x is less that 1, p i=1 f i 2 (x) < p. Then from the General Equivalence Theorem it follows that such a design is both D- and G-optimum. 10. (a). D = 1, 0.2, 0.2, 1. For example ( )/0.5 = 0.2. (b). d(x) = D 2. (c). The maximum of d(x) = 2.92 > 2 (= p), achieved at D = ±1, hence this design is not D-or G-optimum. The design is orthogonal; hence the parameters in the model will be estimated independently from each other. i=1

7 7 Figure 1: G for different values of a (d). D = ±1 are equally good with respect to D-optimality. Non-statistical considerations can be taken into account to decide between the two doses (1 and 2 mg/bw). (e). The new dose must be 3 in order to cover the new required experimental region. The new design is D = 1, 0.6, 0.4, 0, 1. Its standardised variance of prediction is d(x) = 2.156D D The next dose to use is that maximizing d(x). 11. In this case Therefore F = a. G = F T F = 2a 2 + 6, which is maximum when a = ±1 (see Figure 1). Also d(x, ξ) = 3(a 2 2ax x 2 )/(2a 2 + 6). Figure 2 shows plots of d(x) for a = 0, 0.5 and 1. When a = 1, the design is both D- and G-optimum. 12. In this case F = Therefore the standardized variance is [ 1/4 0 d(x) = 4(x 1 x 2 ) 0 1/4. ] ( x1 x 2 ) = x x 2 2

8 8 Figure 2: d(x) for different values of a and has a maximum at x 1 = ±1 and x 2 = ±1 equal to the number of the parameters 2. Hence, the design is D- and G-optimum. Similarly, it is easy to see that a design including the points (1,1) and (-1,1) is both D- and G-optimum as d(x) takes exactly the same form. The optimum design is not unique and alternative symmetric arrangements of the these two points are also D-optimum. Certainly such designs are not useful because one of the explanatory variables is not varied in the design. 13. The General Equivalence Theorem makes it easy to check whether or not an experimental design is D- and G-optimum. It is sufficient to verify that the maximum standardised variance of the prediction is equal to the number of model parameters at the design points, and less elsewhere in the design region. (a) and (b). d(x) = 1 + x 2. The design is D- and G-optimum as x 1. (c) - (e). The designs are not D- and G-optimum: (c) d(x) = x 2 ; (d) d(x) = x 2 ; (e) d(x) = 4(1 x + x 2 )/3. The next point to be added to the design that maximally increases the determinant of the information matrix is -1 or 1 for all designs except for (e), for which it is As defined in Section 10.1, an experimental design ξ is A-optimum if p i=1 λ 1 i is minimized, where λ i is the ith eigenvalue of M(ξ). Using a Lagrange multiplier to constrain the optimization so that p i=1 λ i = p, it becomes clear that the minimum is achieved when all eigenvalues are equal to 1. The D-optimality of any two-level factorial or fractional factorial design used to estimate a first-order polynomial model, which may also include some interactions, was shown in Exercise 9. As M(ξ) = n 1 F T F = I p all its eigenvalues λ i, i = 1,..., p, are equal to 1, hence these designs are also A-optimum. In general an experimental design would be both D- and A-optimum if its information matrix is proportional to the identity matrix.

9 9 15. Orthogonal coding of the list of candidate points ensures that F T c F c = N c I, where F c is the extended design matrix for all candidate points and N c is their number. Minimizing the average standardised variance of prediction over the candidate points is equivalent to minimizing trace[(f T F) 1 ] required by the A-optimality criterion because Nc 1 trace[f c (F T F) 1 F T c ] = N 1 c trace[(f T F) 1 F T c F c ] = trace[(f T F) 1 ], where F is the extended design matrix. 16. Omitting explanatory variables in the analysis of data may lead to wrong results, and is inferior to analysing all data, even if the two sets of variables (those that are considered important and those that are not) are orthogonal. In addition to estimation of the parameters of the linear model, estimation of the error variance is usually important. Omission of important terms in the linear model will lead to heterogeniety of the error variance and underestimation of significance. Similarly, omitting variables in designing an experiment would lead to an inferior design as the estimates of the effects of both types of variable are not independent of each other. In fact, the volume of the confidence ellipsoid for the parameters of interest is proportional to M 11 (ξ) M 12 (ξ)m 1 22 (ξ)m T 12(ξ) = M(ξ) M 22 (ξ), not M 11, provided that the partitioning of the information matrix is done in such a way that the set of the important explanatory variables is first, while the remaining parameters are in the second set. 17. Simplex centroid designs include all possible mixtures in which the components which are included in the mixture have equal proportions. For a Scheffé polynomial of degree d the simplex lattice designs include all possible mixtures in which each component takes values 0, 1/d,..., and 1. In this exercise d = 3. The Simplex Lattice Design and the simplex centroid design for a mixture with components A, B, C and D are given in Tables 2 and 3, respectively. The simplex lattice design is saturated for the Scheffé polynomial of third degree and has 20 observations, while the Simplex Centroid Design has only 15 observation and cannot be used to fit the same model. However, it can be used to estimate the 10 parameters of the second-order Scheffé polynomial model. Hence, if the third-order Scheffé polynomial has to be estimated, the simplex centroid design has to be used. It is recommended to add a few more points to this design in order to test the model for lack of fit. If the observations are collected sequentially, it is recommended that the order of the observations be randomised. 18. (a). In this case F = /2 1/2 0 1/ /2 0 1/2 0 1/ /2 1/ /4,

10 10 Table 2: Simplex Lattice Design A B C D

11 11 Table 3: Simplex Centroid Design A B C D where the first three columns coincide with the design. Then (F T F ) 1 = It is now easy to check that the standardised variance is equal to the number of model parameters (i.e. NN 1 i = 6/1 = 6) at the designs points and less elsewhere. (b). The estimators for the parameters of a second-order Scheffé polynomial are: ˆβ i = y i ˆβ ij = 4y ij 2y i 2y j, where the subscripts i and j correspond to the case when the corresponding mixture components are used. For example, y ij is the response when a mixture of equal amounts of components i and j is used. If replications are available y i and y ij are replaced by the averages of the responses at the corresponding experimental conditions.

12 12 (c). In this case (F T F ) 1 = / The standardised variance at the nonreplicated points is NN 1 i = 8/1 = 8, while at the replicated points it is NN 1 i = 8/3 = You can use the program below to verify this result. proc iml; F = { , , , , , , , }; G = F *F; ginv = inv(g); print ginv; dx = nrow(f) # F * ginv * F ; print dx;} 19. The information matrix of any minimum support design with n i observations at its ith support point is G = F T F = N f T (x i )f(x i ) = i=1 p n i f T (x i )f(x i ) = FS T W F S, where F S is the design matrix for the corresponding saturated design and W is a diagonal matrix with diagonal elements n i, i = 1,..., p. If G is nonsigular the standardized variance of the prediction at the design points i=1 d(x, ξ) = NF T S (F T S W F S ) 1 F S = NW 1 as F S is a square matrix. This completes the proof. 20. (a). Here is an example of a SAS program that can be used to analyse the data. data e20; input dose y; datalines;

13 proc nlin data=e20; parms a = 210 b = 0.1; pred = a*dose/(b+dose); model y = pred; (b). F = [ x 1 β+x 1 ax1 (β+x 1) 2 x 2 β+x 2 ax2 (β+x 2) 2 ]. (c). The result follows from matrix multiplication. (d). As [ G = then and d(x) = [ G 1 = ], x 2 ( x) x2 ( x) x2 ( x) 4. The design is locally D-optimum as the maximum of d(x) is equal to the number of parameters in the model, i.e. 2, and this maximum is reached at the design points (by the General Equivalence Theorem - see Figure 3). (e). When x = β, E(Y ) = α/2, i.e. half of the maximum possible response. This value is often referred to as ED50 or IC50, depending on the nature of the experiment. (f). Similar calculations show that d(x) = x2 ( x) x2 ( x) x 2 ( x) 4. The plot (see Figure 3) suggests that this design is locally D-optimum for the design region where x (0, 1). However, if x > 1 then var(ŷ) exceeds the number of parameters 2 and therefore the design is no longer locally D-optimum. 21. (a). This can be seen by substituting δ = 0 and α = 1 in Model (b). Tedious algebra shows that d(x) = x2 ( ln(x) (ln(x)) 2 ) (x ) 4. ]

14 14 Figure 3: d(x) for the design of Exercise 20(d) Figure 4: d(x) for the design of Exercise 20(f)

15 15 Figure 5: d(x) for the design of Exercise 21(b) Figure 5 illustrates that the maximum of d(x) is equal to the number of the parameters in the model, i.e. 2, and that this maximum is reached at the design points. Hence, by the General Equivalence Theorem, this design is locally D-optimum. The interpretation of β is the same as in Exercise 20.

Statistical Design and Analysis of Experiments Part Two

Statistical Design and Analysis of Experiments Part Two 0.1 Statistical Design and Analysis of Experiments Part Two Lecture notes Fall semester 2007 Henrik Spliid nformatics and Mathematical Modelling Technical University of Denmark List of contents, cont.

More information

The One-Quarter Fraction

The One-Quarter Fraction The One-Quarter Fraction ST 516 Need two generating relations. E.g. a 2 6 2 design, with generating relations I = ABCE and I = BCDF. Product of these is ADEF. Complete defining relation is I = ABCE = BCDF

More information

Fractional Replications

Fractional Replications Chapter 11 Fractional Replications Consider the set up of complete factorial experiment, say k. If there are four factors, then the total number of plots needed to conduct the experiment is 4 = 1. When

More information

FRACTIONAL REPLICATION

FRACTIONAL REPLICATION FRACTIONAL REPLICATION M.L.Agarwal Department of Statistics, University of Delhi, Delhi -. In a factorial experiment, when the number of treatment combinations is very large, it will be beyond the resources

More information

Construction of Mixed-Level Orthogonal Arrays for Testing in Digital Marketing

Construction of Mixed-Level Orthogonal Arrays for Testing in Digital Marketing Construction of Mixed-Level Orthogonal Arrays for Testing in Digital Marketing Vladimir Brayman Webtrends October 19, 2012 Advantages of Conducting Designed Experiments in Digital Marketing Availability

More information

MATH602: APPLIED STATISTICS

MATH602: APPLIED STATISTICS MATH602: APPLIED STATISTICS Dr. Srinivas R. Chakravarthy Department of Science and Mathematics KETTERING UNIVERSITY Flint, MI 48504-4898 Lecture 10 1 FRACTIONAL FACTORIAL DESIGNS Complete factorial designs

More information

CS 147: Computer Systems Performance Analysis

CS 147: Computer Systems Performance Analysis CS 147: Computer Systems Performance Analysis Fractional Factorial Designs CS 147: Computer Systems Performance Analysis Fractional Factorial Designs 1 / 26 Overview Overview Overview Example Preparing

More information

On the Compounds of Hat Matrix for Six-Factor Central Composite Design with Fractional Replicates of the Factorial Portion

On the Compounds of Hat Matrix for Six-Factor Central Composite Design with Fractional Replicates of the Factorial Portion American Journal of Computational and Applied Mathematics 017, 7(4): 95-114 DOI: 10.593/j.ajcam.0170704.0 On the Compounds of Hat Matrix for Six-Factor Central Composite Design with Fractional Replicates

More information

FRACTIONAL FACTORIAL

FRACTIONAL FACTORIAL FRACTIONAL FACTORIAL NURNABI MEHERUL ALAM M.Sc. (Agricultural Statistics), Roll No. 443 I.A.S.R.I, Library Avenue, New Delhi- Chairperson: Dr. P.K. Batra Abstract: Fractional replication can be defined

More information

Minimum Aberration and Related Criteria for Fractional Factorial Designs

Minimum Aberration and Related Criteria for Fractional Factorial Designs Minimum Aberration and Related Criteria for Fractional Factorial Designs Hegang Chen Division of Biostatistics and Bioinformatics 660 West Redwood Street University of Maryland School of Medicine Baltimore,

More information

Lecture 12: 2 k p Fractional Factorial Design

Lecture 12: 2 k p Fractional Factorial Design Lecture 12: 2 k p Fractional Factorial Design Montgomery: Chapter 8 Page 1 Fundamental Principles Regarding Factorial Effects Suppose there are k factors (A,B,...,J,K) in an experiment. All possible factorial

More information

Reference: Chapter 8 of Montgomery (8e)

Reference: Chapter 8 of Montgomery (8e) Reference: Chapter 8 of Montgomery (8e) 69 Maghsoodloo Fractional Factorials (or Replicates) For Base 2 Designs As the number of factors in a 2 k factorial experiment increases, the number of runs (or

More information

TWO-LEVEL FACTORIAL EXPERIMENTS: REGULAR FRACTIONAL FACTORIALS

TWO-LEVEL FACTORIAL EXPERIMENTS: REGULAR FRACTIONAL FACTORIALS STAT 512 2-Level Factorial Experiments: Regular Fractions 1 TWO-LEVEL FACTORIAL EXPERIMENTS: REGULAR FRACTIONAL FACTORIALS Bottom Line: A regular fractional factorial design consists of the treatments

More information

TWO-LEVEL FACTORIAL EXPERIMENTS: IRREGULAR FRACTIONS

TWO-LEVEL FACTORIAL EXPERIMENTS: IRREGULAR FRACTIONS STAT 512 2-Level Factorial Experiments: Irregular Fractions 1 TWO-LEVEL FACTORIAL EXPERIMENTS: IRREGULAR FRACTIONS A major practical weakness of regular fractional factorial designs is that N must be a

More information

3.4. A computer ANOVA output is shown below. Fill in the blanks. You may give bounds on the P-value.

3.4. A computer ANOVA output is shown below. Fill in the blanks. You may give bounds on the P-value. 3.4. A computer ANOVA output is shown below. Fill in the blanks. You may give bounds on the P-value. One-way ANOVA Source DF SS MS F P Factor 3 36.15??? Error??? Total 19 196.04 Completed table is: One-way

More information

THE ROYAL STATISTICAL SOCIETY 2015 EXAMINATIONS SOLUTIONS GRADUATE DIPLOMA MODULE 4

THE ROYAL STATISTICAL SOCIETY 2015 EXAMINATIONS SOLUTIONS GRADUATE DIPLOMA MODULE 4 THE ROYAL STATISTICAL SOCIETY 2015 EXAMINATIONS SOLUTIONS GRADUATE DIPLOMA MODULE 4 The Society is providing these solutions to assist candidates preparing for the examinations in 2017. The solutions are

More information

Strategy of Experimentation III

Strategy of Experimentation III LECTURE 3 Strategy of Experimentation III Comments: Homework 1. Design Resolution A design is of resolution R if no p factor effect is confounded with any other effect containing less than R p factors.

More information

STA 260: Statistics and Probability II

STA 260: Statistics and Probability II Al Nosedal. University of Toronto. Winter 2017 1 Chapter 7. Sampling Distributions and the Central Limit Theorem If you can t explain it simply, you don t understand it well enough Albert Einstein. Theorem

More information

by Christopher Bingham

by Christopher Bingham a D D E!"! #%$&')( $* '$&$,+.-&$0/!12"!"!2 4657 85&$9 ;:=

More information

STAT451/551 Homework#11 Due: April 22, 2014

STAT451/551 Homework#11 Due: April 22, 2014 STAT451/551 Homework#11 Due: April 22, 2014 1. Read Chapter 8.3 8.9. 2. 8.4. SAS code is provided. 3. 8.18. 4. 8.24. 5. 8.45. 376 Chapter 8 Two-Level Fractional Factorial Designs more detail. Sequential

More information

ST3232: Design and Analysis of Experiments

ST3232: Design and Analysis of Experiments Department of Statistics & Applied Probability 2:00-4:00 pm, Monday, April 8, 2013 Lecture 21: Fractional 2 p factorial designs The general principles A full 2 p factorial experiment might not be efficient

More information

ECE 697B (667) Spring 2003

ECE 697B (667) Spring 2003 ECE 697B (667) Spring 2003 Synthesis and Verification of Digital Systems Multi-level Minimization - Algebraic division Outline Division and factorization Definitions Algebraic vs Boolean Algebraic division

More information

A UNIFIED APPROACH TO FACTORIAL DESIGNS WITH RANDOMIZATION RESTRICTIONS

A UNIFIED APPROACH TO FACTORIAL DESIGNS WITH RANDOMIZATION RESTRICTIONS Calcutta Statistical Association Bulletin Vol. 65 (Special 8th Triennial Symposium Proceedings Volume) 2013, Nos. 257-260 A UNIFIED APPROACH TO FACTORIAL DESIGNS WITH RANDOMIZATION RESTRICTIONS PRITAM

More information

Statistica Sinica Preprint No: SS R1

Statistica Sinica Preprint No: SS R1 Statistica Sinica Preprint No: SS-2015-0161R1 Title Generators for Nonregular $2^{k-p}$ Designs Manuscript ID SS-2015-0161R1 URL http://www.stat.sinica.edu.tw/statistica/ DOI 10.5705/ss.202015.0161 Complete

More information

Probability Distribution

Probability Distribution Probability Distribution 1. In scenario 2, the particle size distribution from the mill is: Counts 81

More information

Homework Assignments Sheet. 4) Symbol * beside a question means that a calculator may be used for that question. Chapter 1 Number 9 days

Homework Assignments Sheet. 4) Symbol * beside a question means that a calculator may be used for that question. Chapter 1 Number 9 days Riverside Secondary School Math 10: Foundations and Precalculus Homework Assignments Sheet Note: 1) WS stands for worksheet that will be handed out in class 2) Page numbers refer to the pages in the Workbook

More information

Lecture 14: 2 k p Fractional Factorial Design

Lecture 14: 2 k p Fractional Factorial Design Lecture 14: 2 k p Fractional Factorial Design Montgomery: Chapter 8 1 Lecture 14 Page 1 Fundamental Principles Regarding Factorial Effects Suppose there arek factors (A,B,...,J,K) in an experiment. All

More information

Use of DOE methodology for Investigating Conditions that Influence the Tension in Marine Risers for FPSO Ships

Use of DOE methodology for Investigating Conditions that Influence the Tension in Marine Risers for FPSO Ships 1 st International Structural Specialty Conference 1ère Conférence internationale sur le spécialisée sur le génie des structures Calgary, Alberta, Canada May 23-26, 2006 / 23-26 Mai 2006 Use of DOE methodology

More information

Chapter 30 Design and Analysis of

Chapter 30 Design and Analysis of Chapter 30 Design and Analysis of 2 k DOEs Introduction This chapter describes design alternatives and analysis techniques for conducting a DOE. Tables M1 to M5 in Appendix E can be used to create test

More information

choosedef2(7,4,all:t) K

choosedef2(7,4,all:t) K i!"! a ` a c a ``` `aaa ``` aaa ``` `!ccc j'$k$ 1 C l ; B-?hm 4noqsr $h t=;2 4nXu ED4+* J D98 B v-,/. = $-r

More information

Soo King Lim Figure 1: Figure 2: Figure 3: Figure 4: Figure 5: Figure 6: Figure 7: Figure 8: Figure 9: Figure 10: Figure 11: Figure 12: Figure 13:

Soo King Lim Figure 1: Figure 2: Figure 3: Figure 4: Figure 5: Figure 6: Figure 7: Figure 8: Figure 9: Figure 10: Figure 11: Figure 12: Figure 13: 1.0 ial Experiment Design by Block... 3 1.1 ial Experiment in Incomplete Block... 3 1. ial Experiment with Two Blocks... 3 1.3 ial Experiment with Four Blocks... 5 Example 1... 6.0 Fractional ial Experiment....1

More information

Session 3 Fractional Factorial Designs 4

Session 3 Fractional Factorial Designs 4 Session 3 Fractional Factorial Designs 3 a Modification of a Bearing Example 3. Fractional Factorial Designs Two-level fractional factorial designs Confounding Blocking Two-Level Eight Run Orthogonal Array

More information

TWO-LEVEL FACTORIAL EXPERIMENTS: BLOCKING. Upper-case letters are associated with factors, or regressors of factorial effects, e.g.

TWO-LEVEL FACTORIAL EXPERIMENTS: BLOCKING. Upper-case letters are associated with factors, or regressors of factorial effects, e.g. STAT 512 2-Level Factorial Experiments: Blocking 1 TWO-LEVEL FACTORIAL EXPERIMENTS: BLOCKING Some Traditional Notation: Upper-case letters are associated with factors, or regressors of factorial effects,

More information

Chapter 11: Factorial Designs

Chapter 11: Factorial Designs Chapter : Factorial Designs. Two factor factorial designs ( levels factors ) This situation is similar to the randomized block design from the previous chapter. However, in addition to the effects within

More information

PURPLE COMET MATH MEET April 2012 MIDDLE SCHOOL - SOLUTIONS

PURPLE COMET MATH MEET April 2012 MIDDLE SCHOOL - SOLUTIONS PURPLE COMET MATH MEET April 2012 MIDDLE SCHOOL - SOLUTIONS Copyright c Titu Andreescu and Jonathan Kane Problem 1 Evaluate 5 4 4 3 3 2 2 1 1 0. Answer: 549 The expression equals 625 64 9 2 1 = 549. Problem

More information

Unit 6: Fractional Factorial Experiments at Three Levels

Unit 6: Fractional Factorial Experiments at Three Levels Unit 6: Fractional Factorial Experiments at Three Levels Larger-the-better and smaller-the-better problems. Basic concepts for 3 k full factorial designs. Analysis of 3 k designs using orthogonal components

More information

Confounding and fractional replication in 2 n factorial systems

Confounding and fractional replication in 2 n factorial systems Chapter 20 Confounding and fractional replication in 2 n factorial systems Confounding is a method of designing a factorial experiment that allows incomplete blocks, i.e., blocks of smaller size than the

More information

Suppose we needed four batches of formaldehyde, and coulddoonly4runsperbatch. Thisisthena2 4 factorial in 2 2 blocks.

Suppose we needed four batches of formaldehyde, and coulddoonly4runsperbatch. Thisisthena2 4 factorial in 2 2 blocks. 58 2. 2 factorials in 2 blocks Suppose we needed four batches of formaldehyde, and coulddoonly4runsperbatch. Thisisthena2 4 factorial in 2 2 blocks. Some more algebra: If two effects are confounded with

More information

A General Criterion for Factorial Designs Under Model Uncertainty

A General Criterion for Factorial Designs Under Model Uncertainty A General Criterion for Factorial Designs Under Model Uncertainty Steven Gilmour Queen Mary University of London http://www.maths.qmul.ac.uk/ sgg and Pi-Wen Tsai National Taiwan Normal University Fall

More information

Contents. TAMS38 - Lecture 8 2 k p fractional factorial design. Lecturer: Zhenxia Liu. Example 0 - continued 4. Example 0 - Glazing ceramic 3

Contents. TAMS38 - Lecture 8 2 k p fractional factorial design. Lecturer: Zhenxia Liu. Example 0 - continued 4. Example 0 - Glazing ceramic 3 Contents TAMS38 - Lecture 8 2 k p fractional factorial design Lecturer: Zhenxia Liu Department of Mathematics - Mathematical Statistics Example 0 2 k factorial design with blocking Example 1 2 k p fractional

More information

19. Blocking & confounding

19. Blocking & confounding 146 19. Blocking & confounding Importance of blocking to control nuisance factors - day of week, batch of raw material, etc. Complete Blocks. This is the easy case. Suppose we run a 2 2 factorial experiment,

More information

Design of Experiments

Design of Experiments 6s ANALYZE 6s IMPROVE Chapter 2-4 Design of Experiments 2-4-1 Definition Design of Experiments (DOE) DOE is a systematic approach to investigation of a system or process. A series of structured tests are

More information

Fractional Factorial Designs

Fractional Factorial Designs Fractional Factorial Designs ST 516 Each replicate of a 2 k design requires 2 k runs. E.g. 64 runs for k = 6, or 1024 runs for k = 10. When this is infeasible, we use a fraction of the runs. As a result,

More information

Answers Investigation 3

Answers Investigation 3 Answers Investigation Applications. a., b. s = n c. The numbers seem to be increasing b a greater amount each time. The square number increases b consecutive odd integers:,, 7,, c X X=. a.,,, b., X 7 X=

More information

A case study in Interaction cohomology Oliver Knill, 3/18/2016

A case study in Interaction cohomology Oliver Knill, 3/18/2016 A case study in Interaction cohomology Oliver Knill, 3/18/2016 Simplicial Cohomology Simplicial cohomology is defined by an exterior derivative df (x) = F (dx) on valuation forms F (x) on subgraphs x of

More information

Institutionen för matematik och matematisk statistik Umeå universitet November 7, Inlämningsuppgift 3. Mariam Shirdel

Institutionen för matematik och matematisk statistik Umeå universitet November 7, Inlämningsuppgift 3. Mariam Shirdel Institutionen för matematik och matematisk statistik Umeå universitet November 7, 2011 Inlämningsuppgift 3 Mariam Shirdel (mash0007@student.umu.se) Kvalitetsteknik och försöksplanering, 7.5 hp 1 Uppgift

More information

Fractional Factorials

Fractional Factorials Fractional Factorials Bruce A Craig Department of Statistics Purdue University STAT 514 Topic 26 1 Fractional Factorials Number of runs required for full factorial grows quickly A 2 7 design requires 128

More information

nx + 1 = (n + 1)x 13(n + 1) and nx = (n + 1)x + 27(n + 1).

nx + 1 = (n + 1)x 13(n + 1) and nx = (n + 1)x + 27(n + 1). 1. (Answer: 630) 001 AIME SOLUTIONS Let a represent the tens digit and b the units digit of an integer with the required property. Then 10a + b must be divisible by both a and b. It follows that b must

More information

2 k, 2 k r and 2 k-p Factorial Designs

2 k, 2 k r and 2 k-p Factorial Designs 2 k, 2 k r and 2 k-p Factorial Designs 1 Types of Experimental Designs! Full Factorial Design: " Uses all possible combinations of all levels of all factors. n=3*2*2=12 Too costly! 2 Types of Experimental

More information

SOLUTION. Taken together, the preceding equations imply that ABC DEF by the SSS criterion for triangle congruence.

SOLUTION. Taken together, the preceding equations imply that ABC DEF by the SSS criterion for triangle congruence. 1. [20 points] Suppose that we have ABC and DEF in the Euclidean plane and points G and H on (BC) and (EF) respectively such that ABG DEH and AGC DHF. Prove that ABC DEF. The first congruence assumption

More information

Experimental design (DOE) - Design

Experimental design (DOE) - Design Experimental design (DOE) - Design Menu: QCExpert Experimental Design Design Full Factorial Fract Factorial This module designs a two-level multifactorial orthogonal plan 2 n k and perform its analysis.

More information

Randomisation, Replication, Response Surfaces. and. Rosemary

Randomisation, Replication, Response Surfaces. and. Rosemary Randomisation, Replication, Response Surfaces and Rosemary 1 A.C. Atkinson a.c.atkinson@lse.ac.uk Department of Statistics London School of Economics London WC2A 2AE, UK One joint publication RAB AND ME

More information

Optimal Minimax Controller for Plants with Four Oscillatory Modes Using Gröbner Basis

Optimal Minimax Controller for Plants with Four Oscillatory Modes Using Gröbner Basis 52 ECTI TRANSACTIONS ON ELECTRICAL ENG., ELECTRONICS, AND COMMUNICATIONS VOL.7, NO.1 February 2009 Optimal Minimax Controller for Plants with Four Oscillatory Modes Using Gröbner Basis Chalie Charoenlarpnopparut

More information

β x λ Chapter 11. Supplemental Text Material

β x λ Chapter 11. Supplemental Text Material Chapter. Supplemental Text Material -. The Method of Steepest Ascent The method of steepest ascent can be derived as follows. Suppose that we have fit a firstorder model y = β + β x 0 and we wish to use

More information

Homework 04. , not a , not a 27 3 III III

Homework 04. , not a , not a 27 3 III III Response Surface Methodology, Stat 579 Fall 2014 Homework 04 Name: Answer Key Prof. Erik B. Erhardt Part I. (130 points) I recommend reading through all the parts of the HW (with my adjustments) before

More information

A Survey of Rational Diophantine Sextuples of Low Height

A Survey of Rational Diophantine Sextuples of Low Height A Survey of Rational Diophantine Sextuples of Low Height Philip E Gibbs philegibbs@gmail.com A rational Diophantine m-tuple is a set of m distinct positive rational numbers such that the product of any

More information

Construction of row column factorial designs

Construction of row column factorial designs J. R. Statist. Soc. B (2019) Construction of row column factorial designs J. D. Godolphin University of Surrey, Guildford, UK [Received February 2018. Final revision October 2018] Summary. The arrangement

More information

CS 5014: Research Methods in Computer Science

CS 5014: Research Methods in Computer Science Computer Science Clifford A. Shaffer Department of Computer Science Virginia Tech Blacksburg, Virginia Fall 2010 Copyright c 2010 by Clifford A. Shaffer Computer Science Fall 2010 1 / 254 Experimental

More information

Design of Experiments SUTD - 21/4/2015 1

Design of Experiments SUTD - 21/4/2015 1 Design of Experiments SUTD - 21/4/2015 1 Outline 1. Introduction 2. 2 k Factorial Design Exercise 3. Choice of Sample Size Exercise 4. 2 k p Fractional Factorial Design Exercise 5. Follow-up experimentation

More information

Answers. Investigation 3. ACE Assignment Choices. Applications. = = 210 (Note: students

Answers. Investigation 3. ACE Assignment Choices. Applications. = = 210 (Note: students Answers Investigation ACE Assignment Choices Problem. Core,,, Other Applications ; Connections, ; Etensions 7, ; unassigned choices from previous problems Problem. Core, Other Connections 7; Etensions

More information

CSCI 688 Homework 6. Megan Rose Bryant Department of Mathematics William and Mary

CSCI 688 Homework 6. Megan Rose Bryant Department of Mathematics William and Mary CSCI 688 Homework 6 Megan Rose Bryant Department of Mathematics William and Mary November 12, 2014 7.1 Consider the experiment described in Problem 6.1. Analyze this experiment assuming that each replicate

More information

Computer Aided Construction of Fractional Replicates from Large Factorials. Walter T. Federer Charles E. McCulloch. and. Steve C.

Computer Aided Construction of Fractional Replicates from Large Factorials. Walter T. Federer Charles E. McCulloch. and. Steve C. Computer Aided Construction of Fractional Replicates from Large Factorials by Walter T. Federer Charles E. McCulloch and Steve C. Wang Biometrics Unit and Statistics Center Cornell University Ithaca, NY

More information

INSTITIÚID TEICNEOLAÍOCHTA CHEATHARLACH INSTITUTE OF TECHNOLOGY CARLOW MATRICES

INSTITIÚID TEICNEOLAÍOCHTA CHEATHARLACH INSTITUTE OF TECHNOLOGY CARLOW MATRICES 1 CHAPTER 4 MATRICES 1 INSTITIÚID TEICNEOLAÍOCHTA CHEATHARLACH INSTITUTE OF TECHNOLOGY CARLOW MATRICES 1 Matrices Matrices are of fundamental importance in 2-dimensional and 3-dimensional graphics programming

More information

Strategy of Experimentation II

Strategy of Experimentation II LECTURE 2 Strategy of Experimentation II Comments Computer Code. Last week s homework Interaction plots Helicopter project +1 1 1 +1 [4I 2A 2B 2AB] = [µ 1) µ A µ B µ AB ] +1 +1 1 1 +1 1 +1 1 +1 +1 +1 +1

More information

Fractional Factorial Designs

Fractional Factorial Designs k-p Fractional Factorial Designs Fractional Factorial Designs If we have 7 factors, a 7 factorial design will require 8 experiments How much information can we obtain from fewer experiments, e.g. 7-4 =

More information

The 2 k Factorial Design. Dr. Mohammad Abuhaiba 1

The 2 k Factorial Design. Dr. Mohammad Abuhaiba 1 The 2 k Factorial Design Dr. Mohammad Abuhaiba 1 HoweWork Assignment Due Tuesday 1/6/2010 6.1, 6.2, 6.17, 6.18, 6.19 Dr. Mohammad Abuhaiba 2 Design of Engineering Experiments The 2 k Factorial Design Special

More information

Design of Engineering Experiments Chapter 5 Introduction to Factorials

Design of Engineering Experiments Chapter 5 Introduction to Factorials Design of Engineering Experiments Chapter 5 Introduction to Factorials Text reference, Chapter 5 page 170 General principles of factorial experiments The two-factor factorial with fixed effects The ANOVA

More information

Design of Experiments SUTD 06/04/2016 1

Design of Experiments SUTD 06/04/2016 1 Design of Experiments SUTD 06/04/2016 1 Outline 1. Introduction 2. 2 k Factorial Design 3. Choice of Sample Size 4. 2 k p Fractional Factorial Design 5. Follow-up experimentation (folding over) with factorial

More information

0113ge. Geometry Regents Exam In the diagram below, under which transformation is A B C the image of ABC?

0113ge. Geometry Regents Exam In the diagram below, under which transformation is A B C the image of ABC? 0113ge 1 If MNP VWX and PM is the shortest side of MNP, what is the shortest side of VWX? 1) XV ) WX 3) VW 4) NP 4 In the diagram below, under which transformation is A B C the image of ABC? In circle

More information

Response Surface Methodology

Response Surface Methodology Response Surface Methodology Process and Product Optimization Using Designed Experiments Second Edition RAYMOND H. MYERS Virginia Polytechnic Institute and State University DOUGLAS C. MONTGOMERY Arizona

More information

CS 5014: Research Methods in Computer Science. Experimental Design. Potential Pitfalls. One-Factor (Again) Clifford A. Shaffer.

CS 5014: Research Methods in Computer Science. Experimental Design. Potential Pitfalls. One-Factor (Again) Clifford A. Shaffer. Department of Computer Science Virginia Tech Blacksburg, Virginia Copyright c 2015 by Clifford A. Shaffer Computer Science Title page Computer Science Clifford A. Shaffer Fall 2015 Clifford A. Shaffer

More information

Optimal design of experiments

Optimal design of experiments Optimal design of experiments Session 4: Some theory Peter Goos / 40 Optimal design theory continuous or approximate optimal designs implicitly assume an infinitely large number of observations are available

More information

Design and Analysis of

Design and Analysis of Design and Analysis of Multi-Factored Experiments Module Engineering 7928-2 Two-level Factorial Designs L. M. Lye DOE Course 1 The 2 k Factorial Design Special case of the general factorial design; k factors,

More information

UNIT-8 SIMILAR TRIANGLES Geometry is the right foundation of all painting, I have decided to teach its rudiments and principles to all youngsters eager for art. 1. ABC is a right-angled triangle, right-angled

More information

Resampling and the Bootstrap

Resampling and the Bootstrap Resampling and the Bootstrap Axel Benner Biostatistics, German Cancer Research Center INF 280, D-69120 Heidelberg benner@dkfz.de Resampling and the Bootstrap 2 Topics Estimation and Statistical Testing

More information

A Monte Carlo Evaluation of the Stock Synthesis Assessment Program

A Monte Carlo Evaluation of the Stock Synthesis Assessment Program Fishery Stock Assessment Models 315 Alaska Sea Grant College Program AK-SG-98-01, 1998 A Monte Carlo Evaluation of the Stock Synthesis Assessment Program David B. Sampson and Yanshui Yin Oregon State University,

More information

Unit 5: Fractional Factorial Experiments at Two Levels

Unit 5: Fractional Factorial Experiments at Two Levels Unit 5: Fractional Factorial Experiments at Two Levels Source : Chapter 4 (sections 4.1-4.3, 4.4.1, 4.4.3, 4.5, part of 4.6). Effect aliasing, resolution, minimum aberration criteria. Analysis. Techniques

More information

J JUL - 25-JUL 2016 HOUSEHOLD FINANCES RESEARCH

J JUL - 25-JUL 2016 HOUSEHOLD FINANCES RESEARCH J00 JUL JUL 0 Table XF0 In terms of your finances, how often if at all, do you and your household find yourselves without enough money to buy enough food? BASE: ALL ADULTS AGED + IN GREAT BRITAIN Page

More information

Consider a relation R with attributes ABCDEF GH and functional dependencies S:

Consider a relation R with attributes ABCDEF GH and functional dependencies S: University of Toronto CSC343 Sample 3NF Problem Questions Consider a relation R with attributes ABCDEF GH and functional dependencies S: S = {A CD, ACF G, AD BEF, BCG D, CF AH, CH G, D B, H DEG} 1. Compute

More information

#1, 2, 3ad, 4, 5acd, 6, 7, 8, 9bcd, 10, 11, 12a, 13, 15, 16 #1-5

#1, 2, 3ad, 4, 5acd, 6, 7, 8, 9bcd, 10, 11, 12a, 13, 15, 16 #1-5 MHF4U Unit 3 Rational Functions Section Pages Questions Prereq Skills 146-147 #1, 2, 3bf, 4ac, 6, 7ace, 8cdef, 9bf, 10abe 3.1 153-155 #1ab, 2, 3, 5ad, 6ac, 7cdf, 8, 9, 14* 3.2 164-167 #1ac, 2, 3ab, 4ab,

More information

20g g g Analyze the residuals from this experiment and comment on the model adequacy.

20g g g Analyze the residuals from this experiment and comment on the model adequacy. 3.4. A computer ANOVA output is shown below. Fill in the blanks. You may give bounds on the P-value. One-way ANOVA Source DF SS MS F P Factor 3 36.15??? Error??? Total 19 196.04 3.11. A pharmaceutical

More information

23. Fractional factorials - introduction

23. Fractional factorials - introduction 173 3. Fractional factorials - introduction Consider a 5 factorial. Even without replicates, there are 5 = 3 obs ns required to estimate the effects - 5 main effects, 10 two factor interactions, 10 three

More information

Design of Experiments (DOE) A Valuable Multi-Purpose Methodology

Design of Experiments (DOE) A Valuable Multi-Purpose Methodology Applied Mathematics, 2014, 5, 2120-2129 Published Online July 2014 in SciRes. http://www.scirp.org/journal/am http://dx.doi.org/10.4236/am.2014.514206 Design of Experiments (DOE) A Valuable Multi-Purpose

More information

7. Response Surface Methodology (Ch.10. Regression Modeling Ch. 11. Response Surface Methodology)

7. Response Surface Methodology (Ch.10. Regression Modeling Ch. 11. Response Surface Methodology) 7. Response Surface Methodology (Ch.10. Regression Modeling Ch. 11. Response Surface Methodology) Hae-Jin Choi School of Mechanical Engineering, Chung-Ang University 1 Introduction Response surface methodology,

More information

SHW6-01 Total: 28 marks

SHW6-01 Total: 28 marks SHW6-0 Total: 8 arks Review Exercise 3. Area of C ()(8)sin 9 9.9 BC CA. Let s. 4 9 0 s. Area of C 3. s( s )( s BC)( s CA).(. 4)(. 9)(. 0) 8.0 BAC 80 By the sine forula, BAC 4 80 ( su of ) BC sinc sin A

More information

Semifolding 2 k-p Designs

Semifolding 2 k-p Designs Semifolding 2 k-p Designs Robert W. Mee and Marta Peralta Department of Statistics University of Tennessee Knoxville, TN 37996-0532 rmee@utk.edu Abstract This article addresses the varied possibilities

More information

The Role of Operating Characteristics in Assesing Bayesian Designs in Pharmaceutical Development

The Role of Operating Characteristics in Assesing Bayesian Designs in Pharmaceutical Development The Role of Operating Characteristics in Assesing Bayesian Designs in Pharmaceutical Development Professor Andy Grieve SVP Clinical Trials Methodology Innovation Centre, Aptiv Solutions GmbH Cologne, Germany

More information

The CENTRE for EDUCATION in MATHEMATICS and COMPUTING cemc.uwaterloo.ca Euclid Contest. Wednesday, April 15, 2015

The CENTRE for EDUCATION in MATHEMATICS and COMPUTING cemc.uwaterloo.ca Euclid Contest. Wednesday, April 15, 2015 The CENTRE for EDUCATION in MATHEMATICS and COMPUTING cemc.uwaterloo.ca 015 Euclid Contest Wednesday, April 15, 015 (in North America and South America) Thursday, April 16, 015 (outside of North America

More information

Concurrency and Collinearity

Concurrency and Collinearity Concurrency and Collinearity Victoria Krakovna vkrakovna@gmail.com 1 Elementary Tools Here are some tips for concurrency and collinearity questions: 1. You can often restate a concurrency question as a

More information

Chapter SSM: Linear Algebra Section Fails to be invertible; since det = 6 6 = Invertible; since det = = 2.

Chapter SSM: Linear Algebra Section Fails to be invertible; since det = 6 6 = Invertible; since det = = 2. SSM: Linear Algebra Section 61 61 Chapter 6 1 2 1 Fails to be invertible; since det = 6 6 = 0 3 6 3 5 3 Invertible; since det = 33 35 = 2 7 11 5 Invertible; since det 2 5 7 0 11 7 = 2 11 5 + 0 + 0 0 0

More information

Resampling and the Bootstrap

Resampling and the Bootstrap Resampling and the Bootstrap Axel Benner Biostatistics, German Cancer Research Center INF 280, D-69120 Heidelberg benner@dkfz.de Resampling and the Bootstrap 2 Topics Estimation and Statistical Testing

More information

Statistics for Engineers Lecture 2

Statistics for Engineers Lecture 2 Statistics for Engineers Lecture 2 Antony Lewis http://cosmologist.info/teaching/stat/ Summary from last time Complements Rule: P A c = 1 P(A) Multiplication Rule: P A B = P A P B A = P B P(A B) Special

More information

EXISTENCE AND CONSTRUCTION OF RANDOMIZATION DEFINING CONTRAST SUBSPACES FOR REGULAR FACTORIAL DESIGNS

EXISTENCE AND CONSTRUCTION OF RANDOMIZATION DEFINING CONTRAST SUBSPACES FOR REGULAR FACTORIAL DESIGNS Submitted to the Annals of Statistics EXISTENCE AND CONSTRUCTION OF RANDOMIZATION DEFINING CONTRAST SUBSPACES FOR REGULAR FACTORIAL DESIGNS By Pritam Ranjan, Derek R. Bingham and Angela M. Dean, Acadia

More information

2.830J / 6.780J / ESD.63J Control of Manufacturing Processes (SMA 6303) Spring 2008

2.830J / 6.780J / ESD.63J Control of Manufacturing Processes (SMA 6303) Spring 2008 MIT OpenCourseWare http://ocw.mit.edu 2.830J / 6.780J / ESD.63J Control of Processes (SMA 6303) Spring 2008 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.

More information

Write your Registration Number, Test Centre, Test Code and the Number of this booklet in the appropriate places on the answersheet.

Write your Registration Number, Test Centre, Test Code and the Number of this booklet in the appropriate places on the answersheet. 2016 Booklet No. Test Code : PSA Forenoon Questions : 30 Time : 2 hours Write your Registration Number, Test Centre, Test Code and the Number of this booklet in the appropriate places on the answersheet.

More information

CS 484 Data Mining. Association Rule Mining 2

CS 484 Data Mining. Association Rule Mining 2 CS 484 Data Mining Association Rule Mining 2 Review: Reducing Number of Candidates Apriori principle: If an itemset is frequent, then all of its subsets must also be frequent Apriori principle holds due

More information

Implementing contrasts using SAS Proc GLM is a relatively straightforward process. A SAS Proc GLM contrast statement has the following form:

Implementing contrasts using SAS Proc GLM is a relatively straightforward process. A SAS Proc GLM contrast statement has the following form: Contrasts for Comparison of Means If the analysis of variance produces a significant F-statistic, this signals the need for further analyses. The most common post ANOVA analysis is the comparison of means.

More information

Multilevel Logic Synthesis Algebraic Methods

Multilevel Logic Synthesis Algebraic Methods Multilevel Logic Synthesis Algebraic Methods Logic Circuits Design Seminars WS2010/2011, Lecture 6 Ing. Petr Fišer, Ph.D. Department of Digital Design Faculty of Information Technology Czech Technical

More information

Unit 3: Matrices. Juan Luis Melero and Eduardo Eyras. September 2018

Unit 3: Matrices. Juan Luis Melero and Eduardo Eyras. September 2018 Unit 3: Matrices Juan Luis Melero and Eduardo Eyras September 2018 1 Contents 1 Matrices and operations 4 1.1 Definition of a matrix....................... 4 1.2 Addition and subtraction of matrices..............

More information

Reference: Chapter 6 of Montgomery(8e) Maghsoodloo

Reference: Chapter 6 of Montgomery(8e) Maghsoodloo Reference: Chapter 6 of Montgomery(8e) Maghsoodloo 51 DOE (or DOX) FOR BASE BALANCED FACTORIALS The notation k is used to denote a factorial experiment involving k factors (A, B, C, D,..., K) each at levels.

More information