All methods

MABAC

Available

Multi-Attributive Border Approximation area Comparison. Pamučar and Ćirović (2015)

The idea

One border per criterion
Forklift 2+0.2183Purchase priceWarrantyService networkSpare partsMaintenance costFuel useBearing capacityLifting capacityMovement speedLifting speedForklift 1+0.0826Purchase priceWarrantyService networkSpare partsMaintenance costFuel useBearing capacityLifting capacityMovement speedLifting speedForklift 6+0.0465Purchase priceWarrantyService networkSpare partsMaintenance costFuel useBearing capacityLifting capacityMovement speedLifting speedForklift 7+0.0464Purchase priceWarrantyService networkSpare partsMaintenance costFuel useBearing capacityLifting capacityMovement speedLifting speedForklift 4+0.0246Purchase priceWarrantyService networkSpare partsMaintenance costFuel useBearing capacityLifting capacityMovement speedLifting speedForklift 3-0.0488Purchase priceWarrantyService networkSpare partsMaintenance costFuel useBearing capacityLifting capacityMovement speedLifting speedForklift 5-0.0704Purchase priceWarrantyService networkSpare partsMaintenance costFuel useBearing capacityLifting capacityMovement speedLifting speed

The line in every row is the border G, the geometric mean of the forklifts. Upper area: above it, lower area: below; S adds the sticks up.

Show as table
Alternativeq Purchase priceq Warrantyq Service networkq Spare partsq Maintenance costq Fuel useq Bearing capacityq Lifting capacityq Movement speedq Lifting speedS
Forklift 2+0.0834+0.0960+0.0264-0.0742+0.0455+0.0395+0.0064-0.0160+0.0127-0.0014+0.2183
Forklift 1-0.0372-0.0445-0.0133+0.0468+0.0225+0.0395+0.0190+0.0350+0.0211-0.0062+0.0826
Forklift 6-0.0626+0.0995+0.0264+0.0468-0.0590-0.0615+0.0190+0.0350+0.0044-0.0014+0.0465
Forklift 7+0.0523-0.0270-0.0133+0.0468+0.0560-0.0025-0.0439+0.0180-0.0289-0.0110+0.0464
Forklift 4+0.0406-0.0445-0.0133+0.0468+0.0246-0.0025-0.0187-0.0330-0.0123+0.0370+0.0246
Forklift 3-0.0022+0.0257-0.0530-0.0339-0.0172+0.0059-0.0062+0.0095+0.0211+0.0015-0.0488
Forklift 5-0.0334-0.0445+0.0660-0.0339-0.0381+0.0059+0.0441-0.0245-0.0039-0.0081-0.0704

Same order as a min-max weighted sum

Every dot sits on one line, so no two forklifts can swap; G only decides the sign of S.

Show as table
AlternativeMin-max weighted sumMABAC S
Forklift 20.6899+0.2183
Forklift 10.5542+0.0826
Forklift 60.5181+0.0465
Forklift 70.5180+0.0464
Forklift 40.4962+0.0246
Forklift 30.4228-0.0488
Forklift 50.4012-0.0704

Data: Forklifts (Pamučar and Ćirović 2015).

MABAC draws a border for every criterion, the geometric mean of all alternatives, and adds up how far each alternative lies above or below it.

When to use it

  • Your data mix units or contain zeros or negative values.
  • You want a signed reading: above or below the border.
  • Results should compare with DEMATEL, Entropy or CRITIC then MABAC studies.

Algorithm

  1. 1Min-max normalization

    The worst value of each column becomes 0 and the best 1, so zeros and negative values work too.

  2. 2Weighted matrix

    The +1 keeps every value positive, so the geometric mean in the next step is defined.

  3. 3Border approximation area

    Computed as exp of the mean log, which stays exact where the product would underflow (hundreds of alternatives).

  4. 4Distance from the border

    Positive: upper area; negative: lower area.

  5. 5Score

    Rank by decreasing S: the min-max weighted sum minus one constant, so the border only sets the sign.

Used together with

Why these pairs
  • DEMATEL then MABAC: Pamučar and Ćirović (2015), the forklift example used here.
  • Entropy then MABAC: Biswas and Das (2018), nine hybrid vehicles, a test example.
  • CRITIC then MABAC: Çelik (2020), Turkish participation banks, a test example end to end.
  • The origin compares MABAC with SAW, COPRAS, TOPSIS, MOORA and VIKOR.

Reference example

Pamučar, D.; Ćirović, G. (2015)Matches the printed digits
Details

Pamučar, D.; Ćirović, G. (2015). The selection of transport and handling resources in logistics centers using Multi-Attributive Border Approximation area Comparison (MABAC). Expert Systems with Applications 42(6), 3016-3028. doi:10.1016/j.eswa.2014.11.057

Published example, reproduced by the tests: seven forklifts, ten criteria (three costs), DEMATEL weights (Tables 5, 7-12). G and S match the printed 4 decimals (difference below 0.0001) and all seven ranks match. Also tested: Biswas and Das (2018, nine hybrid vehicles, all 9 printed digits) and Çelik (2020, six Turkish participation banks, CRITIC weights computed here), all ranks.

Checked by this site's test suite.

Sources

Show 5 sources
  1. Pamučar and Ćirović (2015). The selection of transport and handling resources in logistics centers using Multi-Attributive Border Approximation area Comparison (MABAC). Expert Systems with Applications 42(6), 3016-3028. doi:10.1016/j.eswa.2014.11.057
  2. Pamučar & Ćirović (2015), Expert Systems with Applications 42(6), 3016-3028 doi:10.1016/j.eswa.2014.11.057
  3. Biswas & Das (2018), Management Science Letters 8(2), 121-130 doi:10.5267/j.msl.2017.11.004
  4. Çelik (2020), CRITIC-MABAC, İslam Ekonomisi ve Finansı Dergisi 6(2), 311-334 dergipark.org.tr/tr/download/article-file/1194510
  5. pyDecision 5.1.8, mabac_method (ignores the weights; not a reference) github.com/Valdecy/pyDecision