Klaus D. Goepel, BPMSG, Contact, last update: Sep 4, 2017
393 thoughts on “Feedback”
OOPS its Belinda again, I went to the feedback page before the download page. I have now downloaded the corrected version. thank you. By the way the AHP is being used for Masters Research to weight different threats to biodiversity in a spatial conservation planning context.
Hi Belinda,
thanks for your feedback. I don’t have a specific preference for the citation. Could be:
Klaus D. Goepel, AHP excel template, Oct 2012, Business Performance Management Singapore, http://bpmsg.com
Dear Mr Goepel
Thank you for this fantastic template, it is very convenient for non-mathemetically minded people like myself. I have used your AHPcalc ver. 22.5.12. Could you please check the summary sheet after the input for 7 participants – if I select participant 1,2 or 3 their individual results are displayed. From participant 4 onwards (4,5,6,7) aswell as the consolidated geometric mean, the same result is returned. Can I assume that the result being returned is the correct consolidated result, and that there is not a calculation error as such but rather an error in the display of individual results for participants 4 through 7. The consolidated result CR is 3.9% which is very low compared to the 7 inputs (which range from 7% to 31%). I hope I am making sense. I look forward to your response.
Specifically, I plugged-in the numbers from Table 3 in your spreadsheet. The results from your spreadsheet were close but not the same. Since I don’t see your calculations in the spreadsheet, I can’t tell why …
Also compared to an example in Table 2 in Saaty’s paper in Int. J. Services Sciences, Vol. 1, No. 1, 2008 and againl the results are close but not quite the same…
Hi Mihail.
thanks for your feedback. The deviations are a result of numerical inaccuracy due to excel and the algorithm. Calculating the Eigenvector I use the power method with 6 iterations, and standard excel matrix operations. In my template you see all steps in worksheet ‘8×8. In this sheet you may also check the validity of the iterated Eigenvector, as I am doing the reverse calculation using the Eigenwert equation
(A – lambda * I)* Eigenvector = 0
In your example given, the result is not exactly 0 for all matrix elements (some are 0.1)
I had a note in my previous AHP download page on this, but I’ll make a new post to highlight it. Overall – from a practical point of view – it does not really has a big impact, as the judgment scale 1-9 is rough and subjective anyway.
Thanks, Klaus
Hi Klaus,
Your software is a great help. It makes my work really fast.
I did not understand the concept of (6) iterations. The books used in MBA classes do not discuss this concept; may be because they explain how to do AHP manually.
Could you please explain the concept of iterations. Why it is needed; should we always use normalized principal eigenvector values that appear on the summary page of excel file. Any other source of help would be great.
I thank you in advance and look forward.
Hi Asif,
thanks for the feedback. The iteration is necessary to calculate the Eigenvector. It is one of the possible algorithm for the numerical computation. In principle you have to iterate until the Eigenvector is “stable”. From my experience with AHP I found out that with the given number of criteria (3 to 8 max), usually 6 iterations are sufficient. That’s the reason you find 6 iterations in my excel sheet.
Yes, use always the values appearing on the summary sheet.
Here 2 links (from wikipedia) to pdf documents with more explanations:
Thanks Klaus for your answer.
I have two points to discuss:
First, I guess the AHP template makes use of geometric mean of the inputs. So we do not need to calculate geometric mean. Is it correct?
Second, there are multiple methods for calculating eigenvector. One such method is through taking nth root. How does the concept of iteration apply in such case?
The concept of iterations is clear when using your method of ‘raising power till eigenvector is stable’ But what about using the other methods. Please see page 10 of the following link for eigenvector calculation through nth root: http://www.booksites.net/download/coyle/student_files/AHP_Technique.pdf
Regards, Asif
Hi Asif,
Yes, you are right; I use the geometric mean.
Regarding the calculation: I didn’t really look at other methods, so difficult to answer, which one is the best, or how many iterations are needed for the root algorithm.
Thanks for the link.
Regards, Klaus
You can unprotect the input sheets, no password required. Then change the data validation of the corresponding cells. (standard excel functionality)
Regards, Klaus
Dear sir,
I had done my experimental work on machining of steel. The variable are Current(3-levels), voltage(2-levels) ,polarity(2-levels) , electrode. The out put is Metal removal rate, tool wear rate. Can you please help me to set the AHP diagram to select the optimum solution
Thanks
Hi,
sorry, but I don’t really have enough free time to help you on this. I even think, AHP might not be the optimal method. Have a look at the NIST Handbook engineering statistics. You find the link on my website under external links. They have a chapter dealing with experiments exactly like yours.
Regards, Klaus
Hi Klaus,
Great tool–thanks so much for sharing! I noticed that on the Input sheets, in cells b63-b66, the formulas have an error. Instead of pointing to J1, they increment to J2, J3, etc. Small thing, easily fixed 🙂
Good day my name is Stan, i am a novice in the use of AHP, please help. Worked on your latest spreadsheet that is AHP calc version 7.4.12 example of “Finding priority of product feature” with the following criteria color, memory and screen resolution but with no alternatives.
I have already attempted to use your newly spreadsheet but with alternatives. Using the same example but now selecting a better computer with two alternatives (i.e Sarah Computer and Magic Computer) but using the given criteria’s (color, memory and screen resolution). So need to compare the criteria of the two computers to arrive at buying the best computer. Thank you.
Dear Klaus,
My name is Amir, I’m MBA student and working on my final project which is related to ANP. First of all I would like to appreciate your effort to share your precious understanding and experiences in this area. Second, I have a question about entering the data into the Super Decision software. My problem is that I can only enter the first questionnaire’s data! I don’t know how to enter the next questionnaires’ data! Actually there is no place for entering the data for more than one questionnaire! It is highly appreciated if you can guide me with this issue.
Dear Amir,
You are right, the template is only for one questionnaire. If you want to use more comparisons, you have to fill out the matrix manually. There is one sheet prepared, where you can input the comparisons as element of the matrix and calculate the geometric mean for each element. But it requires manual change of formulas and references. Sorry, didn’t develop it further.
Regards, Klaus
Dear Klaus,
As I terminate my final project which was dealing with ANP-based approach, I would like to thank you again for your efforts in terms of illustrating and clarifying the vague areas of super decision software. I also want to request your e-mail address if you don’t mind. I leave my e-mail here so you can contact me with that if you don’t tend to put your e-mail address here. I look forward to hearing from you soon.
My name is Amir too, I’m Industrial management student and working on my final project which is related to ANP, like you. I have a same problem “inserting more than one questionnaire data in Suoer Decision.
unfortunately I didn’t understand the Klaus’s answer. can you help me?
I thank you in advance and look forward…
Regards, Amir
OOPS its Belinda again, I went to the feedback page before the download page. I have now downloaded the corrected version. thank you. By the way the AHP is being used for Masters Research to weight different threats to biodiversity in a spatial conservation planning context.
Is there specific citation you prefer us to use.
Regards
Hi Belinda,
thanks for your feedback. I don’t have a specific preference for the citation. Could be:
Klaus D. Goepel, AHP excel template, Oct 2012, Business Performance Management Singapore, http://bpmsg.com
regards, Klaus
Dear Mr Goepel
Thank you for this fantastic template, it is very convenient for non-mathemetically minded people like myself. I have used your AHPcalc ver. 22.5.12. Could you please check the summary sheet after the input for 7 participants – if I select participant 1,2 or 3 their individual results are displayed. From participant 4 onwards (4,5,6,7) aswell as the consolidated geometric mean, the same result is returned. Can I assume that the result being returned is the correct consolidated result, and that there is not a calculation error as such but rather an error in the display of individual results for participants 4 through 7. The consolidated result CR is 3.9% which is very low compared to the 7 inputs (which range from 7% to 31%). I hope I am making sense. I look forward to your response.
Belinda
Hi,
Just stumbled on your YouTube videos and from there, to your Excel spreadsheet. I compared it to this paper (you can download the full paper from that side): http://www.slideshare.net/ricardo.vargas/using-the-analytic-hierarchy-process-ahp-to-select-and-prioritize-projects-in-a-portfolio
Specifically, I plugged-in the numbers from Table 3 in your spreadsheet. The results from your spreadsheet were close but not the same. Since I don’t see your calculations in the spreadsheet, I can’t tell why …
Also compared to an example in Table 2 in Saaty’s paper in Int. J. Services Sciences, Vol. 1, No. 1, 2008 and againl the results are close but not quite the same…
Thanks for the feedback!
Hi Mihail.
thanks for your feedback. The deviations are a result of numerical inaccuracy due to excel and the algorithm. Calculating the Eigenvector I use the power method with 6 iterations, and standard excel matrix operations. In my template you see all steps in worksheet ‘8×8. In this sheet you may also check the validity of the iterated Eigenvector, as I am doing the reverse calculation using the Eigenwert equation
(A – lambda * I)* Eigenvector = 0
In your example given, the result is not exactly 0 for all matrix elements (some are 0.1)
I had a note in my previous AHP download page on this, but I’ll make a new post to highlight it. Overall – from a practical point of view – it does not really has a big impact, as the judgment scale 1-9 is rough and subjective anyway.
Thanks, Klaus
Mihail,
now my new version (11.12.12) is exactly reproducing the results in in both papers!
Hi Klaus,
Your software is a great help. It makes my work really fast.
I did not understand the concept of (6) iterations. The books used in MBA classes do not discuss this concept; may be because they explain how to do AHP manually.
Could you please explain the concept of iterations. Why it is needed; should we always use normalized principal eigenvector values that appear on the summary page of excel file. Any other source of help would be great.
I thank you in advance and look forward.
Hi Asif,
thanks for the feedback. The iteration is necessary to calculate the Eigenvector. It is one of the possible algorithm for the numerical computation. In principle you have to iterate until the Eigenvector is “stable”. From my experience with AHP I found out that with the given number of criteria (3 to 8 max), usually 6 iterations are sufficient. That’s the reason you find 6 iterations in my excel sheet.
Yes, use always the values appearing on the summary sheet.
Here 2 links (from wikipedia) to pdf documents with more explanations:
http://www.boku.ac.at/mi/ahp/ahptutorial.pdf
http://www.business.pitt.edu/faculty/papers/saaty-into-to-ahp-mathematics.pdf
Regards, Klaus
Thanks Klaus for your answer.
I have two points to discuss:
First, I guess the AHP template makes use of geometric mean of the inputs. So we do not need to calculate geometric mean. Is it correct?
Second, there are multiple methods for calculating eigenvector. One such method is through taking nth root. How does the concept of iteration apply in such case?
The concept of iterations is clear when using your method of ‘raising power till eigenvector is stable’ But what about using the other methods. Please see page 10 of the following link for eigenvector calculation through nth root:
http://www.booksites.net/download/coyle/student_files/AHP_Technique.pdf
Regards, Asif
Hi Asif,
Yes, you are right; I use the geometric mean.
Regarding the calculation: I didn’t really look at other methods, so difficult to answer, which one is the best, or how many iterations are needed for the root algorithm.
Thanks for the link.
Regards, Klaus
Hi,
How to put a value like 0,2 in the Input 1, intensity?
Do I have a way to unlock that?
Best regards.
You can unprotect the input sheets, no password required. Then change the data validation of the corresponding cells. (standard excel functionality)
Regards, Klaus
Dear sir,
I had done my experimental work on machining of steel. The variable are Current(3-levels), voltage(2-levels) ,polarity(2-levels) , electrode. The out put is Metal removal rate, tool wear rate. Can you please help me to set the AHP diagram to select the optimum solution
Thanks
Hi,
sorry, but I don’t really have enough free time to help you on this. I even think, AHP might not be the optimal method. Have a look at the NIST Handbook engineering statistics. You find the link on my website under external links. They have a chapter dealing with experiments exactly like yours.
Regards, Klaus
Klaus :
Thanks very much for sharing your program. I will use it for teaching AHP to graduate students.
Best regards,
Jaime chandia
Hi Klaus,
Great tool–thanks so much for sharing! I noticed that on the Input sheets, in cells b63-b66, the formulas have an error. Instead of pointing to J1, they increment to J2, J3, etc. Small thing, easily fixed 🙂
Best regards,
Stephen
Hi Stephen,
thank you so much for your valuable feedback. I have corrected the version and updated for download on the download page
regards, Klaus
Good day my name is Stan, i am a novice in the use of AHP, please help. Worked on your latest spreadsheet that is AHP calc version 7.4.12 example of “Finding priority of product feature” with the following criteria color, memory and screen resolution but with no alternatives.
I have already attempted to use your newly spreadsheet but with alternatives. Using the same example but now selecting a better computer with two alternatives (i.e Sarah Computer and Magic Computer) but using the given criteria’s (color, memory and screen resolution). So need to compare the criteria of the two computers to arrive at buying the best computer. Thank you.
Respectfully submitted
Stan
Dear Stan,
thanks for your feedback. Please see my reply from Feb 2012 to the similar question from Tara here: http://bpmsg.com/ahp-excel-template/
Dear Klaus,
My name is Amir, I’m MBA student and working on my final project which is related to ANP. First of all I would like to appreciate your effort to share your precious understanding and experiences in this area. Second, I have a question about entering the data into the Super Decision software. My problem is that I can only enter the first questionnaire’s data! I don’t know how to enter the next questionnaires’ data! Actually there is no place for entering the data for more than one questionnaire! It is highly appreciated if you can guide me with this issue.
Best regards,
Amir
Dear Amir,
You are right, the template is only for one questionnaire. If you want to use more comparisons, you have to fill out the matrix manually. There is one sheet prepared, where you can input the comparisons as element of the matrix and calculate the geometric mean for each element. But it requires manual change of formulas and references. Sorry, didn’t develop it further.
Regards, Klaus
Dear Klaus,
As I terminate my final project which was dealing with ANP-based approach, I would like to thank you again for your efforts in terms of illustrating and clarifying the vague areas of super decision software. I also want to request your e-mail address if you don’t mind. I leave my e-mail here so you can contact me with that if you don’t tend to put your e-mail address here. I look forward to hearing from you soon.
Regards,
Amir
My name is Amir too, I’m Industrial management student and working on my final project which is related to ANP, like you. I have a same problem “inserting more than one questionnaire data in Suoer Decision.
unfortunately I didn’t understand the Klaus’s answer. can you help me?
I thank you in advance and look forward…
Regards, Amir