why it is being run mostly by distance, and is part-time. It is for practi- tioners who can try out
ideas that they gain, from the course, from this
text-book and from their peers working with them in workshops during
the programme.
In the well known trad
The Business School brought together representatives from both
the Management Consultancies Association (MCA) and the
Institute of Management Consultants (IMC), with a representative
from Kogan Page (the Publishers) and academics from the
University of S
require very different competences on the part of the consultant. Thus,
the consultant will sometimes adopt the role of mentor, at other times the
role of creator, while on other occasions the consultant is, in reality, a
leader.
Consultants have also bee
theory comes out of best practice and that best practice can be generated
and enhanced when practitioners are prepared to try out theoretical
ideas to see if they actually work.
Of course this book also stands alone. It can and should be read and
used by
true or not, it does underline one truism unless client and consultancy
work together on a constant basis, the likelihood of real added value
resulting is debatable.
This book deals with the nuts and bolts of management consultancy.
This is important, but
one of the camps then those who are not with you are not deemed
management consultants. As a student of management consultancy you
are faced, therefore, with a dilemma before your studies have even
begun. But do not worry, management consultancy is all ab
60 billion in value. Having said this, over 80 per cent of management
consultancy is still performed nationally, and even in a global market it
is fairer to say that it is sold internationally but still performed nation- ally. It is fair to add that at
th
the consultancy trade associations. It concludes by returning to the pre- sent and suggesting
where this industry is heading.
Definitions of consultancy
There are almost as many definitions of consultancy as there are consul- tants; each consultant
and co
Page 16 of 491
Foreword
I welcome this opportunity to write the Foreword to this invaluable
book that succinctly describes the management consultancy industry.
Indeed I would congratulate the authors on even attempting to do so,
for today the spectrum cov
Now we are facing a second evolution. As clients increasingly under- stand not only the value of
outside bought in advice, they also, through
their regular use of management consultancy, understand the service
that is on offer. Management consultancy nowa
whole occupation, including the Management Consultancy companies,
and so he has regarded the occupation as an industry. Of course he is also
right this is one of the interesting things about different ways of looking
at the same occupation different inter
debate as to whether Management Consultancy is an industry or a pro- fession. Of course this
debate will continue, but the real concern of
Management Consultants should perhaps be about their own profes- sionalism, irrespective of
the status of the occupa
and y, respectively, then in the future (as we select other axes), the distance Dcurrent(c;d) should be
related to the given distance function D by D2current(c;d)=D2(c;d)(xy)2 The explanation is suggested by
Fig. 22. Here, then, is the outline of the Fast
choiceof clusters is reconsidered if there are too many to
t the representations in main memory, or if a cluster gets too big (too high a radius). There are many
details about what happens when new points are added. Here are some of the key points:
(a) A
.8 Processing a Main-Memory-Full of Points in BFR
With the
rst load of main memory, BFR selects the k cluster centroids, using some chosen main-memory
algorithm, e.g., pick a sample of points, optimize the clusters exactly, and chose their centroids as th
dean space. It also assumes that there is too much data to
t in main memory. The data structure it uses to store clusters is like an R-tree. Nodes of the tree are disk
blocks, and we store di
erent things at leaf and interior nodes:
_ In leaf blocks, we s
idean space with one dimension for each position, we can de
ne the distance function D(x;y)=jxj+jyj2jLCS(x;y)j, where LCS stands for the longest common subsequence
of x and y. In our example, LCS(abcde;bcdxye) isbcde, of length 4, so D(abcde;bcdxye)=5+62_
lled \discarded" points, these points in truth have a signi
cant e
ect throughout the running of the algorithm, since they determine collectively where the centroid is and
what the standard deviation of the cluster is in each dimension.
28
2
1
Figure 17:
rly, k = 3 is the right number of clusters, but suppose we
rst try k = 1. Then all points are in one cluster, and the average distance to the centroid will be high.
x x x xxxx x
xx x x x
x x xx x x
radius
k 123
Average
4
Figure 16: Discovering that k = 3
the pointindicatedas a. Suppose that when we assign 4, we
nd that 4 is closer to 2 than to a, so 4 joins 2 in its cluster, whose center thus moves to b. Finally, 5 is
closer to a than b, so it joins the cluster f1;3g, whose centroidmovesto c. 2 _ Wecanini
es indicate the associated time for each state. 2
ABABC
000000
11111
222
3
2
Figure 33: Maintaining the set of states and their associated times
10.6 CountingCompositeEvents
Toextendtheaboveideastocompositeepisodes,wemustkeepa\machine"ofsomesortforeach su
hose distance from Q is the minimum, where \distance" is de
ned by the \energy" of the di
erence of the sequences; i.e., D(S;T)=R1 0S(t)T(t)_2dt.Forinstance,the Si's might be records of the prices
of various stocks, and Q is the price of IBM stock, delaye
isode. It is the serial composition of three episodes. The
rst is the single event A. Then comes the parallel epsiode fB;C;Dg. The third is a composite episode
consisting of the parallel composition of the serial episodes (E;F) and (G;H). Two examples of
le the spring of length 5 is stretched to 17/3. In this con
guration, the total energy of the system is (3 7 3)2 + (4 10 3 )2 + (5 17 3 )2 =4 =3. 2
31
7/3 10/3
17/3
a
b
cba c
4
3
5
Figure 20: Optimal placement of three points in one dimension
8.2 Fastmap
increase k to 4, then one of the true clusters will be art
cially parttoned into two nearby clusters, as suggested by the solid lines. The average distance to
centroid will drop a bit, but not much. It is this failure to drop further that tps us o
that k
checkserialepisode(A1;A2;:;An)wesimulateanondeterministic
niteautomatonthatrecognizes the string :_A1A2_An, as suggested by Fig. 31. As we scan the events in
the data, we keep track of the set of states that the NFA is in.
.
any
q q q q AAA 12 012 n n
any
oved slightly closer to the mean, as follows:
(a) Pick the
rst sample point to be the point of the cluster farthest from the centroid. (b) Repeatedly pick additional
sample points by choosing that point of the cluster whose minimum distance to an already
t)e2_jitdt. Recall that the imaginary exponential eix is de
ned to be sinx + icosx. Thus, the real and imaginary parts of Xj tell how well S(t) matches sine and cosine
functions that have j periods within the interval 0-to-1, i.e., sin2_jt and cos2_jt.
Ex
hose distance from query Q is no more than _2, compute the
rst m Fouriercoe_cientsof Q, and look for points in the space no more distant than _ from the point
corresponding to Q.
4. Since there are false drops, compare each retrieved sequence with Q to be
CHAPTER 8
Bond Valuation and the Structure of Interest Rates
Learning Objectives
1. Explain what an efficient capital market is and why market efficiency is important to
financial managers.
2. Describe the market for corporate bonds and three types of cor