noc19-mg15 Lec 30-ELECTRE

IIT Kanpur July 201
20 followers
Dec 3, 2023
Simple ELECTRE method, epsilon ELECTRE method, concordance dominance matrix, discordance dominance matrix
[Music]
[Music]
welcome back my dear friends and dear
students a very good morning good
afternoon good evening to all of you
wherever you are at whatever time and
this is the D a DM - which is data
analysis and decision making to course
on the NPTEL book series and as you know
this course is for 12 weeks which is
spread over sixty lectures and the total
duration in hours if you calculate is 30
hours and each week we have five
lectures so and which which each lecture
being for half an hour and after each
week we have assignments to be solved by
the students and as you know after this
course there would be and some
examination and my name is Ragunan and
Sengupta from the Ayami department IIT
Kanpur in India so this if you if you
see the slide this is the DDM lecture
number 30 which is the sixth week end of
the sixth eighth classes so we will try
to wrap up the elektra process in the
electro process initially we are
discussed about General Directorate then
we are basically talking about epsilon
electron and in the epsilon vector if
you remember I have been discussing that
how you can make your decision choice in
such a way that you have the discordant
set concordant set and the indifferent
set and depending on the liking
disliking I will basically divide my
whole choice joy my means the decision
maker choice can be divided into two
three mutually exclusive and exhaustive
sets where disk concordance means when I
take decision a K with respect to L and
J being the criteria if I am willing to
take that I get some positive points
hence I will basically put that value of
J into the concordance set or C and an
ultimate link comes into the those
values that will be utilized to find out
the matrix capital C and if if I am able
to take a decision a key with respect to
Al but in the case the criteria J
doesn't give me that benefit so those
index value or indices values which
doesn't support a K will be utilized to
calculate the D matrix which is the
Gordon matrix and in the epsilon method
we added an extra set of decision-making
tool or or a rule for decision-making
where that in different set would also
become covered where we are whatever
decision we take with respect to J if J
means the criteria if a K and al doesn't
give me any benefit by taking makin a k
i'll put that j into the set which is i
in in different indices values which
will be utilized finally to calculate
the indifference matrix and if we
remember i did also mention in the last
class which was in the 29th class that
the indifference matrix or an inference
set can be a symmetric values in the
sense that we can utilize the asymmetric
penalty loss functions based on the
civil engineering problem the marketing
the marketing problem and the electoral
engine problem according these where we
can being some concept of practicality
into the decision process so continuing
our discussion if you remember we were
doing the values calculating the values
of C 1 1 D 1 1 and I 1 1 where C D I are
basically the the values in the
concordance set discordant set and the
indifference set similarly we will
calculate C 1 to C 1 the one after that
will consider C 1 to C D 1 2 and I 1 2
then going to C 1 3 D 1 3 and I 1 3 so
we'll go column wise or row wise so here
we are trying to calculate C 2 1 and D 2
1 and I 2 1 corresponding to the fact
that these are as I mentioned I am again
repeating concordance concept this court
in concept in different concept so we
take again the values of epsilon epsilon
means the difference which you have as
zero point I'll highlight it as 0.01 for
all the cases I'm not changing it even
though I did mention that you can
consider depending on the
a value the epsilon value can change but
I am NOT considering that for J is equal
to 1 will have the values compared the
y2 1 with respect to Y 1 1 and we will
see that I 1 which is G is equal to on
Falls in the concordance at similarly
for J is equal to 2 2 falls in the
discordant set and G is equal to 3 falls
in the indifferent set so we have for C
2 1 D 2 1 I 2 1 the values as marked in
the yellow highlight point then when we
go into C 2 2 D 2 2 I 2 2 obviously here
the comparison would be company amongst
themselves so obviously all of them
would fall into the set of indifferent
because I am now indifferent whether I
like it or not like it doesn't matter
now because I'm indifferent between
trying to compare one with itself like
comparing the one with 1 or 2 it 2 or 3
with 3 but here we are doing the 2 comma
to sell values obviously in that case
and the null set would be seen all set
would be D and I set would have all the
elements which is I J I am J is equal to
1 G is equal to 2 and J is equal to 3
now we still go to the next level would
be C 2 2 C 2 3 D 2 3 I 2 3 then we'll go
to C 3 1 D 3 1 and 9 3 1 so C 2 3 D 2 3
and I 2 3 would be the so-called values
in 2 comma 3 cell in the whole overall
matrix where each cell if you consider
the 3 by 3 matrix over the cell would
have elements corresponding to C D n I
so the 1 comma 1 would be C 1 1 D 1 1 I
1 1 then 2 comma 2 would be C 2 2 D 2 2
I 2 2 and the 3 comma 3 cell in that
matrix would be C 3 3 D 3 3 and I 3 3 so
when you are considering 2 3 it will be
2 comma 3 cells position so
corresponding to J is equal to 1 2 3
which are the criteria if you again come
compare again considering the fact that
we are taking epsilon as 0.01 which is
just an assumption simplistic assumption
we have that the value again ok now here
I'll try to use the different colors
in the indifference that it would be
yellow in the concordance that it will
be green and in the discordance that it
will be red so as it is in different set
one falls in in different death set 3
falls in in different set so 3 is 1 and
3
now the correspondingly in in discordant
set two Falls there so 2 becomes red in
color and the green one is a null set
because C 2 3 is null hence it falls in
the null element is there in C 2 3 when
I go to 3 comma 1 cells that means hence
we have to consider the concordance
discordance and the different sets as C
3 1 d 3 1 i 3 1 so if we consider again
j is equal to 1 2 3 so J 3 are the
criteria it can be more than that also
corresponding to that we have 1 2 3 all
falling in the in differences obviously
they would be yellow so so the in
difference set has all the elements 1 2
3 and the corresponding D set is null
corresponding C set is normal so we are
trying to now differentiate if you
remember that a the real line trying to
divide into three zones one in between
zone D will depend on the epsilon value
would be the yellow one where I mean
different on to the right when I am
looking from my side would be the green
one which is concordance positive one
and on to the left of that indifference
or node with the red one which is the
discordant set now I go to the cell
number 3 comma 2 in that overall matrix
so again so I'll consider C 3 2 D 3 2
and I 3 2 and corresponding to J is
equal to 1 2 3 when I calculate I am NOT
going to the detail collection you can
basically check and we are considering
epsilon value as 0.01 for all the three
cases G is equal to 1 G is equal to V J
is equal to 3 all that criteria you are
not changing them so based on that when
we have we
we would have G is equal to one is
indifferent G is equal to three is
equally indifferent and if I come to the
concordance values it is two is equal to
the concordance set so correspondingly
c-32 is element only one element which
is two that is j is equal to two if i
consider red it is which is the
discordant is a null set and if i say
consider yellow which is the indifferent
set would be 1 & 3 in J is equal to 1 J
is equal to 3 now I come to the last
cell which is 3 comma 3 then you will
have basically the con column the
concordance discordance and the in
different concepts would be C 3 3 D 3 3
and I 3 3 the indices similarly for J is
equal to 1 2 3 considering epsilon as
0.01 using the same calculation which
has shown will basically have all of
them falling into the set C so they
would be 1 2 3 falling into which my my
mistake I think I have done it something
wrong let me change it sorry
just change it here so they would be
technically 1 it's my typo error sorry
for that
extremely sorry so once I have this let
me save it so I'll again highlight the
appropriate colors for the cell 3 comma
3 so I will have corresponding to I so
this is 1 is equal to 1 3 2 is equal to
1 3 I I equal to 1 2 3 3 3 3 is equal to
thin I 3 3 2 is equal to I 3 3 1 is
going to I 3 3 so the in different cell
has all the elements 1 2 3 and the
corresponding cells or the element in C
3 3 is null column in 7 D 3 3 is null so
you are formulated the rule so now if
you see the relative structure of the
construct which are formed so
1 comma 1 2 comma 2 & 3 comma 3 all the
values I'll use the yellow I'll use the
C for example the violet just marker so
the indifference set in the principal
diagonal are all the elements so liking
disliking and that is liking concordance
and disliking means tea discordant sets
are null so in the principal diagonal
everything when you compare you are
indifferent in the decision and if you
take the concepts of the of the diagonal
element then you will basically have the
indifferent and this concordance set and
the in this discordant set values based
on the which you will proceed remember
that this actual value of C this capital
C which you have written is nothing to
do with Concord it's just a symbolic
representation of that matrix C so this
should technically be a symmetric in
nature
now consider the weights are given so
what I do is that our what we considered
that we consider different weights for
the concordance at different weights for
the in different set in different
weights for the disc ordinances so
weights are considered in a way such
that I will assign 1/3 1/3 1/3 for the
case for concordance and that means the
the top rows values would be 1/3 for the
1 cross 2 and corresponding with the
diagonal name in along the principal
diagonal the opposite mirror image which
is 2 cross 1 will be 1/3 and if you
consider here 1/3 and the opposite L
means - it means the liking disliking
concept has been formulated accordingly
similarly for the in difference one it
is mirror not a mirror image the
principal diagonal is 1 now when I
compare 1/2 to 1 I am indifferent when I
compared to 3 and 3 - it is indifferent
but when I compare 1 3 and 3 1 then
obviously this value of 1 and -
basically gives the indifference concept
Martineau starkly similarly I have the
discordance weights and based on that
you what what is important to note is
that find out if you find out the sum of
all the corresponding cells for this
weighted or width matrix for CDI
it would be 1 and that can be that will
give you an idea that based on the
weights you are doing the know you have
done the normalization in the right way
so if we end up
see for example first element - second
element 1 13 1 - so is 1 if we add up
one-third one-third one-third it is 1 if
you have - - 1 its 1 so if you had a
one-third one-third one-third it is 1
then if we add - one - is 1 and if we
had - 2/3 1/3 is 1 so lastly going to
the last row - - - okay you have to
check this there must be some L double
double checked it so it should be
definitely be 1 here also my mistake so
verify that two-third 130 now so this
should be so if I add up 1 so now it
should be - one - is 1 then 1/3 2/3 - is
1 and - one - is one so all these values
comes out to one so in this principal
diagonal value wise the symmetricity is
maintained in the indifference part not
in the case for the the discordant set
and the concordant set the values now if
you follow the rule again we will just
follow the same concept what we have
done for the discord in fact like we
give a range and try to my find out the
maximum value based on all the
combinations so if maniac trying to
compare will compare one cell by cell
value with respect to all the
combinations which will be there along
the row all the quality
pending on which way you have
normalization you've taken place so
you'll basically come find out the small
C value small I value small D value when
you are comparing the Kate and I let
this decisions based on the criteria J
so you'll basically compare them for all
all the criterias and basically take the
max value or the ratios accordingly so
if I do that so I'll just go if you
remember the values the cells let me
write it down so it's easy for you so
this would be good good point let me
take these cells values so if I so sorry
I'm not skipping but I'm trying to bring
that slide in the picture so it will be
easier for us to compare these please
bear with me because I thought initially
then because it was so I'll remove this
and this this Y values would be nucleus
accordingly so I'll use this time and
again so I am copying it for our own
benefit so we have these values of Y so
it will be now much easier for us to
denote yes so for all the students and
the readers I'll basically request them
to have a look at this matrix which was
why why was mystically the normalized
using the utility sense and normalizing
along the row or the column the X matrix
multiplied the weights we had that Y
values and based on that we are trying
to proceed you have already found out
the contours and discordant concept
so giving the confidence and discordance
I find out the cell values so c1 1 then
C 2 2 and C 3 C the principal diagonal
will be - - - because concordance
discordant concept doesn't come here so
if I use these values so c1 2 becomes 1
c1 3 becomes 0 just pause and see so you
will basically take the maximum of these
values accordingly so once we have that
I'll just proceed accordingly so then
once c2 1 will be point 5 5 6 7 c2 3
would be 0 so if you check here 1 2 + 1
3 are the placements of the cells which
you are talking about 1 comma 2 - 1
comma 3 similarly when you come to 2
comma 1 2 comma 2 were obviously will be
dead - 1 2 comma 3 so corresponding
values of C 2 comma 1 would be 0.5 5 6 7
and use the same formula which we have
just discussed that for the C value I
value and the D value the small C ID
which I have just mentioned and
utilizing the Y matrix normalized values
so C 2 3 comes out to be 0 similarly 3
comma 1 which is C 3 1 would be 0 C 3 2
would be 1 now you have found out the C
value similarly you have to find out the
small I value simulator to find out that
D small D values and the calculations
would be given here and just go through
the final results so once you have that
the C matrix comes out to be here this
is basically the concordance so called
matrix which you have the principal
diagonal which I mentioned is - and of
the diagonal elements have been given so
only part is to require is if you
compare 0.5 5 6 7 with 1 it means that
if I take 1 with respect to 2 I get much
more benefit then when I take 2 with
respect to 1 that means for the
conglomeration of the count or the
collective positive points which I get
for choosing 1 with respect to 2 for all
the criterias taken together I get a
much higher benefit
similarly when I come to I 1 1 I 2 2 and
I 3 3 which is the principle diagonal in
the capital I matrix it this is the
indifference matrix it will be - and of
the diagonal elements are calculated
here so 1 I 1 2 is 0.1 1 I 1 3 is 1
similarly when I go to I 2 1 and I 2 3
the corresponding values are 0.1 1 and
point 2 8 6 3 I am going a little bit
fast because we take the formula they
just a deputation just put them and
solve it accordingly
similarly i3 1 + i3 2 values are 1 and
0.286 3 so based on this when I find out
the indifferent matrix the well value
comes out to be like this so this would
be a symmetric matrix because in the
indifferent matrix here the concept of
symmetry city would hold true not for
concordance and discordance control is a
liking and disliking concepts are little
bit different so this is and use the
highlighter this is symmetric this is
symmetric and this is symmetric so that
was not the case when we consider the C
matrix and we will also see may not be
or may not hold true for the D matrix
similarly when I come to the D matrix
which is the discordance actual matrix
which you have so similarly D 1 1 D 2 2
and D 3 3 the principal diagonal will be
- values because you are not comparing
them and then off the diagonal values
are given as d 1 2 is 0.55 671 3 is 0
when I go into D 2 1 and D 2 3 the
values are 1 respectively in both the
cases and when I go to D 3 1 and D 3 to
both the values are 0 so utilizing this
when I have the mid D matrix so you
check so the principal diagonal values
are not symmetric and this is very
important this is a metric that's
coincident but this is also not
symmetric
or I should use a different colors so
this is also not symmetric and this try
to be seen I'm using a different color
in order to this is also this is
symmetric while other elements are not
symmetric now I need to find out the CC
star I star and D star values so these
are the average values which I find out
the formulas which I use depending on M
into 1 by M into M minus 1 the values
which you have the values which you
calculate for the unconfident
discordance dominance matrix and the
indifference matrix which you want to
find out C star comes out to be 0.16 6 7
I start point
triple 7 8 and again this should be D
star so this value comes out of the D
star also comes out to be point 1 6 6 7
similarly on based on this I need to
find out the capital F matrix capital G
matrix and then multiply the
corresponding elements for F and G then
find it out so the Concordance concept
discordant concept when the in
difference concept for the liking in
disliking are if you remember in the
general director process what basically
subsumed under the matrix F which is a
set when I take the positive decision G
was the set when I take the negative
decision and then I try to compare them
and find out that if I multiply those
values whether they are positive or 0 so
once I have the F set for C I and D so
the values are coming out to be these
remember for the principal diagonal they
may be symmetric definitely symmetric
for the I in difference concept but not
for the DNC so if you check so and
this this color scheme has nothing to do
so this is symmetric and here indeed
this is symmetric if I consider this one
this is a symmetric in the D set or this
D matrix this is a symmetric but if I
consider the values for I symmetric
symmetric symmetric so I would continue
remaining the symmetric one and not for
the D or C it is a it may be just a
coincidence if C and D are symmetric but
it would not be so when I go and try to
find out the G capital matrix when I am
taking the other decision so what is the
disk benefit so correspondingly again
the values of C I and D concordance
average values discordance average
values in difference on average values
they come out to be point four two six
one point six five is four six five four
and point four two six one again so
based on that once I formulate the G
matrix for C I and D again you will see
this values which I have and use the
color for blue blue so this is symmetric
which is fine this is symmetric for I
matrix everything is symmetric remember
and this is also symmetric here but for
C and D as there is symmetricity
and a symmetricity these are the case
and once you have them you will
basically want to find out the combined
effect like if I have taken K when I try
to find out the benefit I find and would
break down the benefit to see set I set
D set and find out the total benefit and
then try to find out the so-called dis
benefit
again I am basically divided into CDN I
combine them and basically at the end of
the day I have an F matrix for CDI a G
matrix for CDI and can multiply the
corresponding cells of C with C C with
DC with I the next would be in order to
find out the ranking then I compare I
with C I with I and I with D and finally
I'll compare the concept of D D with C D
with I and D with D so once I have this
so I'll try to find need to find out the
e matrix only remember the subscripts I
have not gone gone with the detailed
calculation we need to find out ECC that
means concordance concordance ECI
concordance in in different ECD
concordance and and this concept of
discordance so I will multiply the
corresponding values from for the cset
with the cset transferring to F and G
then C set with I set for F and G and C
set for deeds d set with F ng
F and G will remain the same only the
substrate starts changing when I go to
the in different set again I only have
basically IC iin ID that is in different
concordance in different in different in
different discordance the value is
multiplied F and G but the subscripts
are IG I see III and ID and finally when
I'd find out the discordance so called
overall matrix it will be e DC di D D
which discordance concordance
discordance in different discordant
discordance the values multiplied would
be from the FN and G in the matrix the
corresponding subsidies and TD so with
this I will end and the 30th lecture
which has to do with with Elektra
process both the Elektra and normal
process then and the epsilon later and
we'll consider the other topics of
topsis and other matters in the sub
second class starting the next week have
a nice day and thank you very much for
your attention
[Music]
[Music]
[Music]
