All methods

WASPAS

Available

Weighted Aggregated Sum Product ASsessment. Zavadskas et al. (2012)

The idea

Sum or product: λ sets the blend

Each line is one alternative; where two lines cross, they swap places. The vertical line marks λ = 0.5.

Show as table
Alternativeλ = 0.0λ = 0.1λ = 0.2λ = 0.3λ = 0.4λ = 0.5λ = 0.6λ = 0.7λ = 0.8λ = 0.9λ = 1.0
A10.79750.79900.80050.80200.80350.80500.80650.80800.80950.81100.8125
A20.75330.75760.76200.76630.77070.77510.77940.78380.78810.79250.7969
A30.76470.76580.76690.76800.76910.77020.77130.77240.77340.77450.7756
A40.78740.78920.79100.79280.79460.79640.79820.80000.80180.80360.8055
A50.76740.77160.77580.77990.78410.78820.79240.79660.80070.80490.8091

Data: Textile suppliers (Aytaç Adalı and Tuş Işık 2017).

WASPAS scores each alternative as a weighted sum and as a weighted product, then blends the two with λ (1 = sum only, 0 = product only).

When to use it

  • Every value is a positive measurement or score.
  • You want a blend of compensating and weakness-sensitive scoring.
  • You want to see whether the order holds as λ moves.

Algorithm

  1. 1Linear max normalization
  2. 2Weighted sum and weighted product
  3. 3Blend

    Rank by decreasing Q; λ = 1 is SAW and λ = 0 the weighted product model.

Used together with

Why these pairs
  • The origin example takes its weights from Entropy (Zavadskas et al. 2012).
  • Chakraborty & Zavadskas (2014) use AHP weights in their cutting-fluid and industrial-robot examples.
  • SWARA then WASPAS for supplier selection in Aytaç Adalı & Tuş Işık (2017), the example used here.
  • At λ = 1 WASPAS is SAW with linear max normalization; at λ = 0 it is the weighted product model.

Reference example

Aytaç Adalı, E.; Tuş Işık, A. (2017)Matches the printed digits
Details

Aytaç Adalı, E.; Tuş Işık, A. (2017). Bir Tedarikçi Seçim Problemi İçin SWARA ve WASPAS Yöntemlerine Dayanan Karar Verme Yaklaşımı. International Review of Economics and Management 5(4), 56-77. doi:10.18825/iremjournal.335408

Published example, reproduced by the tests: five suppliers, six criteria (three of them cost), SWARA weights (Tables 4-7). The normalized matrix of Table 6 matches to 3 decimals and all 55 scores of Table 7 (λ = 0, 0.1, ..., 1) to 4 decimals; the three rank orders the paper reports over λ are the same. Two more examples are tested: Chakraborty & Zavadskas (2014), industrial robots (Tables 13-15), and the origin (Zavadskas et al. 2012, Table 2), whose Table 1 prints one value of a1 as 1.0000 where every result needs 0.1000; its Table 4 also prints a1's optimal λ as 0.32 for 0.34, and its last column header reads λ = 10 for 1.

Checked by this site's test suite.

Sources

Show 5 sources
  1. Zavadskas et al. (2012). Optimization of Weighted Aggregated Sum Product Assessment. Elektronika ir Elektrotechnika 122(6), 3-6. doi:10.5755/j01.eee.122.6.1810
  2. Zavadskas, Turskis, Antucheviciene & Zakarevicius (2012), Elektronika ir Elektrotechnika 122(6), 3-6 doi:10.5755/j01.eee.122.6.1810
  3. Chakraborty & Zavadskas (2014), Applications of WASPAS method in manufacturing decision making, Informatica 25(1), 1-20 doi:10.15388/Informatica.2014.01
  4. Aytaç Adalı & Tuş Işık (2017), SWARA-WASPAS supplier selection, International Review of Economics and Management 5(4), 56-77 doi:10.18825/iremjournal.335408
  5. pyDecision 5.1.1, waspas_method (1 + min-max normalization; not the origin) github.com/Valdecy/pyDecision