Consumers in a certain state choose between three long distance telephone services. GTT, NCJ, and Dash. Aggressive marketing by all three companies results in a continual shift of customers among the three services. Each year, GTT loses 25% of its customers to NCJ and 20% to Dash, NCJ loses 5% of its customers to GTT and 20% to Dash, and Dash loses 15% of its customers to GTT and 5 % to NCJ. Assuming that these percentages remain valid over a long period of time, what is each company's expected market share in the long run? Please help me solve this.
Every year, GTT: 5% to NCJ, 20% to Dash NCJ: 25% to GTT, 25% to Dash Dash: 30% to GTT, 30% to NCJ Now let G, N, D be the proportion of market share of GTT, NCJ, and Dash respectively. We want G = G + 0.25N + 0.3D - 0.05G - 0.2G = 0.75G + 0.25N + 0.3D N = N + 0.05G + 0.3D - 0.25N - 0.25N = 0.5N + 0.05G + 0.3D D = D + 0.2G + 0.25N - 0.3D - 0.3D = 0.4D + 0.2G + 0.25N G + N + D = 1 Simplifying, 0.25G = 0.25N + 0.3D 0.5N = 0.05G + 0.3D 0.6D = 0.2G + 0.25N G + N + D = 1 Intuitively, the 4 equations imply that the net change in market share for each company must be zero. Solving, G = 52.63% N = 21.05% D = 26.32%
Does this help?
these numbers are different in my problem. that is the wrong answer
@mathmate
@Hero @Mehek14
@Kainui
What course is this? This looks like a markov matrix, and the long-term behavior is given by the eigenvector associated with lambda = 1
you should get 20% , 30% , 50% for G, N and D
@phi that was the answer!!!!! i actually have the same problem with different numbers if you can help me again.... Consumers in a certain state choose between three long distance telephone services. GTT, NCJ, and Dash. Aggressive marketing by all three companies results in a continual shift of customers among the three services. Each year, GTT loses 10% of its customers to NCJ and 25% to Dash, NCJ loses 10% of its customers to GTT and 30% to Dash, and Dash loses 30% of its customers to GTT and 15 % to NCJ. Assuming that these percentages remain valid over a long period of time, what is each company's expected market share in the long run?
you should first find the markov matrix, so that \[\left[\begin{matrix}G \\ N \\ D\end{matrix}\right]=A\left[\begin{matrix}G \\ N \\ D\end{matrix}\right]\]
can you do that?
no I'm lost with this. i have asked my professor but haven't heard from him
how much does G lose (in total) to its competitors?
35%?
yes, so if it gained nothing from the other 2, how much would it have the next year ?
in other words, what percent does it keep?
65%
yes, so it would get G_new= 0.65 G_old
\[\left[\begin{matrix}G \\ N \\ D\end{matrix}\right]=\left[\begin{matrix}0.65 & ?& ? \\ ? & ?&?\\ ? & ?&?\end{matrix}\right]\left[\begin{matrix}G \\ N \\ D\end{matrix}\right]\]
do you know the G on the left side is calculated by multiplying the top row of the matrix times the vector on the right?
.65 .10 .30 .10 .60 .15 .25 .30 .55
i didnt know that. i don't know how to find the vector on the right
as a check, the columns should sum to 1 and they do. next, we have to find the eigenvector for eigenvalue 1 do you know how to do that ?
would it be -.35 .10 .30 .10 -.40 .15 .25 .30 -.45
yes, that is a good start A- I (in this case) now find reduced row echelon form
this is where i got stuck the last time ... how do i find the form
step 1: use elimination to zero out the first column can you do that ?
do i divide the first row by 1/65?
-.35 .10 .30 .10 -.40 .15 .25 .30 -.45 you multiply the first row by 0.10/0.35 and add the resultant to the 2nd row
.025? i added it to the entire row
the idea is 0.10/0.35 * -0.35 = -0.10 and when you add that to the 2nd row, the first entry is zeroed out
0.10/0.35 is 0.28571 when you multiply the first row by that you get -0.100000 0.028571 0.085714 when you add that row to the 2nd row you get 0.00000 -0.37143 0.23571
ok i understand now.... what is the next step
you use the same idea to zero the first entry in the last row
how did you get -0.37143 0.23571 in the second row?
starting at the beginning multiply the first row by 0.10/0.35 you get -0.100000 0.028571 0.085714
you then add that to the 2nd row -0.100000 0.028571 0.085714 .10 -.40 .15
ok then i know you add -0.100000 to the first one in the second row giving you 0
you get a new 2nd row 0.00000 -0.37143 0.23571
ok .. i will try the 3rd row now
-0.100000 0.028571 0.085714 0.00000 -0.37143 0.23571 .15 0.328571 -.364286
is that correct?
you leave the first row as it was. The -0.100000 0.028571 0.085714 is only a "side calculation" in other words, after we zero out the first entry in the 2nd row *only* the 2nd row in the matrix changes. when you do the 3rd row, you still use the *original* first row.
here is what you have after doing the first row, but before doing the 3rd row -.35 .10 .30 0.00000 -0.37143 0.23571 .25 .30 -.45
*after doing the 2nd row, but before doing the 3rd row
yes i forgot you multiply ... i was adding what is the next step
zero out the 0.25 in the 3rd row you do that by multiplying the first row by 0.25/0.35 to get a *temporary row" that you add to the 3rd row to get a *new third row*
is the third row .... 0 .37142857 -0.23571429
yes. good news, it looks close to the 2nd row
ok what is the second row
what is the next step***
-.35 .10 .30 0.00000 -0.37143 0.23571 .25 .30 -.45 becomes -.35 .10 .30 0.00000 -0.37143 0.23571 0 .37142857 -0.23571429
yes i have that
we go to the next column over, and down 1 row the new "pivot" is the -0.37143 multiply the 2nd row by 1 (i.e. do nothing) and add it to the 3rd row. This will eliminate the entry below the pivot
you will get -.35 .10 .30 0.00000 -0.37143 0.23571 0.000 0.00000 0.0000
next, we want the pivots to be 1 the first pivot is the -0.35 in location (1,1) to make it one, we divide the 1st row (all entries) by -0.35 to get a *new first row*
1 .28571427 .85714286 0.00000 -0.37143 0.23571 0.000 0.00000 0.0000
is that correct?
except for the signs. they should be negative for the other 2 entries, right ?
correct i was diving by .35 not -.35
dividing*
1 - 0.28571427 - 0.85714286 0.00000 -0.37143 0.23571 0.000 0.00000 0.0000 now make the 2nd pivot 1. that is, divide the 2nd row by -0.37143 to get a new 2nd row
1 - 0.28571427 - 0.85714286 -0.37143 1 -0.63460141 0.000 0.00000 0.0000
the first entry in the 2nd row stays 0
oh ok
1 - 0.28571427 - 0.85714286 0 1 -0.63460141 0.000 0.00000 0.0000
finally, we want the pivot columns to be all zeros (except for the pivot) in other words, we want to "clear out" the -0.28... above the 2nd pivot to do that, multiply the 2nd row by 0.2857... and add that to the first row to get a *new 1st row* (the 2nd row stays as is)
0.2857 0 0.67582818 0 1 -0.63460141 0.000 0.00000 0.0000
you did something wrong. the 2nd row (temp result) should be 0 0.28571427 -0.1813... now add that to the 1st row 1 - 0.28571427 - 0.85714286
oh ok
what is the matrix ?
1 0 - 1.03544286 -0.37143 1 -0.63460141 0.000 0.00000 0.0000
sorry this.... 1 0 - 1.03544286 0 1 -0.63460141 0.000 0.00000 0.0000
better!
now we "read off" the eigenvector or we solve for the eigenvector
how?
do you know how to find vectors in the "null space" of a matrix? (or have you heard of this?)
sorry i have not
not to get to far off track: the eigenvector equation is \[ A x = \lambda x \\ Ax - \lambda x = 0 \\ \left( A - \lambda I\right) x = 0 \] and x is a vector that gives 0 (i.e. a vector of zeros) when we multiply by the matrix \( \left( A - \lambda I\right) \)
all of which means reduced row echelon form matrix we found times the the eigenvector x gives 0 \[\left[\begin{matrix}1 & 0 & a \\ 0 & 1 & b \\ 0 & 0 & 0\end{matrix}\right]\left[\begin{matrix}x \\ y \\z\end{matrix}\right]=\left[\begin{matrix}0\\0 \\0\end{matrix}\right]\]
the z (associated with the row of zeros) is a free variable. People usually pick it to have a value of 1 now solve for x and y (I used a and b instead of the long decimals)
remember the matrix is short-hand for a system of equations in this case 0*z = 0 (let z=1 , though any value would work, 1 is a nice choice) 0x + 1y + b*z= 0 1 x + 0*y +a*z = 0 solve for y and x
you get x= -a, y=-b, z = 1 in other words, the vector is \[\left[\begin{matrix}-a \\ -b \\ 1\end{matrix}\right]\]
where in our case, a and b are the entries in the last column of the reduced row echelon matrix
ok sorry I'm trying to do it on my end .... I'm lost
the last step is we want the entries in the vector to sum up to 1 so divide the entries by the sum
what step ?
so x= -1.03544286 y= -.63460141 z= 1
almost. the equation (for x for example) is x -1.035*z = 0 where we let z= 1 we get x - 1.035 = 0
in other words, we negate the two entries in the 3rd column when we use them in the eigenvector
ok so how do we divide by the sum
my assignment is do in a few so I'm trying to understand this quicker
add up the entries [ 1.03 0.63 1] (you can use more decimals) then divide each entry by that value
and, if you like, multiply that vector by 100, so the values represent percent
isn't the value 1? if we divide by 1 we will get the same answer??
1.03846 0.63462 1.00000 does not add up to 1 but we want it to. so divide each entry by the sum 2.6731
it's the same idea as changing [ 1 2] into ⅓ ⅔
Join our real-time social learning platform and learn together with your friends!