All methods

AHP

Available

Analytic Hierarchy Process. Saaty (1977)

The idea

A matrix of ratios, one vector that fits it
FlexibilityOpportunitiesOpportun…SecurityReputationReputati…SalaryWeightFlexibility11/40.301/60.141/40.481/80.070.0361Opportunities411/30.4731.641/70.240.1219Security63143.511/20.520.2619Reputation41/31/411/70.150.0746Salary872710.5056Pairwise ratios a_ik · Fitted w_i / w_kLargest gap: Flexibility / Reputation

Upper cells hold the comparisons, lower cells their reciprocals, and the bars on the right show the eigenvector weights (CR 0.073, threshold 0.10).

Show as table
CriterionFlexibilityOpportunitiesSecurityReputationSalaryWeight
Flexibility11/41/61/41/80.0361
Opportunities411/331/70.1219
Security63141/20.2619
Reputation41/31/411/70.0746
Salary872710.5056

Data: Choosing a job (Saaty 2008).

AHP turns pairwise judgements ("how much more important is A than B?", 1‑9) into weights and a consistency ratio CR.

When to use it

  • Importance comes from people.
  • Up to about 9 criteria: n(n-1)/2 comparisons, 36 at n = 9.
  • You want a check of how consistent the judgements are.

Algorithm

  1. 1Pairwise comparison matrix

    Only the upper triangle is asked; the rest follows from reciprocity.

  2. 2Priority vector (eigenvector)

    w is the principal right eigenvector, found by power iteration: multiply by A and renormalize until the change is below 1e-12.

  3. 3Geometric-mean variant (option)

    For n = 3 it gives the same weights as the eigenvector; for n of 4 or more they differ slightly.

  4. 4Consistency index and ratio

    Random index RI: n = 3: 0.58, 4: 0.90, 5: 1.12, 6: 1.24, 7: 1.32, 8: 1.41, 9: 1.45, 10: 1.49, 11: 1.51, 12: 1.48, 13: 1.56, 14: 1.57, 15: 1.59, and the Alonso & Lamata (2006) linear fit above 15; for n = 1 or 2, CR is not applicable.

  5. 5Alternative local priorities and distributive synthesis

    When AHP also ranks the alternatives, each criterion j yields local priorities p_ij (from an m x m pairwise matrix or from sum-normalized table shares, with 1/x for cost) and distributive synthesis combines them into global priorities P_i.

Used together with

Why these pairs
  • AHP then TOPSIS is the most common pipeline: 10,451 title or abstract co-mentions in OpenAlex, and TOPSIS is AHP's first partner every year in the review by Çebi et al. (2022).
  • AHP then VIKOR (1,753 co-mentions), for example Kaya & Kahraman (2010) on energy planning in Istanbul.
  • AHP then PROMETHEE (1,484 co-mentions), for example Dağdeviren (2008) on equipment selection.
  • AHP then ELECTRE (817 co-mentions), for example Soner & Önüt (2006) on supplier selection.
  • Roszkowska (2013) compares AHP weights with the rank formulas ROC, RS and RR for five criteria.

Reference example

Saaty, T.L. (2008)Matches the printed digits
Details

Saaty, T.L. (2008). Decision making with the analytic hierarchy process. International Journal of Services Sciences 1(1), 83-98. doi:10.1504/IJSSCI.2008.017590

Eigenvector weights match Saaty (2008) Table 3 (job criteria, 5 x 5) and Table 4 (3 x 3) to the printed 3 decimals. The drinks matrix of Table 2 (7 x 7) is not reciprocal as printed (Tea/Wine = 2 but Wine/Tea = 1/3); with Tea/Wine = 3 every published value and CR = 0.022 are reproduced, so the 2 is a typo. Also checked against Doğuç (2022) Tables 4-7 and the incomplete-matrix LLSM example of Bozóki, Fülöp & Rónyai (2010) Example 2.

Checked by this site's test suite.

Sources

Show 10 sources
  1. Saaty (1977). A scaling method for priorities in hierarchical structures. Journal of Mathematical Psychology 15(3), 234-281. doi:10.1016/0022-2496(77)90033-5The book is Saaty (1980); the algorithm and the consistency rule follow Saaty (2008) and Saaty (1988), the geometric-mean variant Crawford & Williams (1985).
  2. Saaty (1977), Journal of Mathematical Psychology 15(3), 234-281 doi:10.1016/0022-2496(77)90033-5
  3. Saaty (2008), International Journal of Services Sciences 1(1), 83-98 doi:10.1504/IJSSCI.2008.017590
  4. Crawford & Williams (1985), Journal of Mathematical Psychology 29(4), 387-405 doi:10.1016/0022-2496(85)90002-1
  5. Alonso & Lamata (2006), International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems 14(4), 445-459 doi:10.1142/S0218488506004114
  6. Bozóki, Fülöp & Rónyai (2010), Mathematical and Computer Modelling 52, 318-333 doi:10.1016/j.mcm.2010.02.047
  7. Doğuç (2022), Ekonomi, İşletme ve Maliye Araştırmaları Dergisi 4(2), 126-145 doi:10.38009/ekimad.1142952
  8. Saaty (1988), How to make a decision: the analytic hierarchy process, ISAHP proceedings
  9. Roszkowska (2013), Optimum. Studia Ekonomiczne 5(65), 14-33 doi:10.15290/ose.2013.05.65.02
  10. Çebi, Onar, Öztayşi & Kahraman (2022), Integration of AHP with other MCDM methods: a literature review, ISAHP 2022 isahp.org/uploads/36_001.pdf