A basic, but very reasonable, assumption is that if attribute A is absolutely more
important than attribute B and is rated at 9, then B must be absolutely less important
than A and is valued at 1/9.
These pairwise comparisons are carried out for all factors to be considered, usually not
more than 7, and the matrix is completed. The matrix is of a very particular form which
neatly supports the calculations which then ensue (Saaty was a very distinguished
mathematician).
The next step is the calculation of a list of the relative weights, importance, or value, of
the factors, such as cost and operability, which are relevant to the problem in question
(technically, this list is called an eigenvector). If, perhaps, cost is very much more
important than operability, then, on a simple interpretation, the cheap equipment is
called for though, as we shall see, matters are not so straightforward. The final stage is
to calculate a Consistency Ratio (CR) to measure how consistent the judgements have
been relative to large samples of purely random judgements. If the CR is much in
excess of 0.1 the judgements are untrustworthy because they are too close for comfort to
randomness and the exercise is valueless or must be repeated. It is easy to make a
minimum number of judgements after which the rest can be calculated to enforce a
perhaps unrealistically perfect consistency.
The AHP is sometimes sadly misused and the analysis stops with the calculation of the
eigenvector from the pairwise comparisons of relative importance (sometimes without
even computing the CR!) but the AHP’s true subtlety lies in the fact that it is, as its
name says, a Hierarchy process. The first eigenvector has given the relative importance
attached to requirements, such as cost and reliability, but different machines contribute
to differing extents to the satisfaction of those requirements. Thus, subsequent matrices
can be developed to show how X, Y and Z respectively satisfy the needs of the firm.
(The matrices from this lower level in the hierarchy will each have their own
A basic, but very reasonable, assumption is that if attribute A is absolutely more
important than attribute B and is rated at 9, then B must be absolutely less important
than A and is valued at 1/9.
These pairwise comparisons are carried out for all factors to be considered, usually not
more than 7, and the matrix is completed. The matrix is of a very particular form which
neatly supports the calculations which then ensue (Saaty was a very distinguished
mathematician).
The next step is the calculation of a list of the relative weights, importance, or value, of
the factors, such as cost and operability, which are relevant to the problem in question
(technically, this list is called an eigenvector). If, perhaps, cost is very much more
important than operability, then, on a simple interpretation, the cheap equipment is
called for though, as we shall see, matters are not so straightforward. The final stage is
to calculate a Consistency Ratio (CR) to measure how consistent the judgements have
been relative to large samples of purely random judgements. If the CR is much in
excess of 0.1 the judgements are untrustworthy because they are too close for comfort to
randomness and the exercise is valueless or must be repeated. It is easy to make a
minimum number of judgements after which the rest can be calculated to enforce a
perhaps unrealistically perfect consistency.
The AHP is sometimes sadly misused and the analysis stops with the calculation of the
eigenvector from the pairwise comparisons of relative importance (sometimes without
even computing the CR!) but the AHP’s true subtlety lies in the fact that it is, as its
name says, a Hierarchy process. The first eigenvector has given the relative importance
attached to requirements, such as cost and reliability, but different machines contribute
to differing extents to the satisfaction of those requirements. Thus, subsequent matrices
can be developed to show how X, Y and Z respectively satisfy the needs of the firm.
(The matrices from this lower level in the hierarchy will each have their own
正在翻譯中..