All methods

CILOS

Available

Criterion Impact LOSs. Zavadskas and Podvezko (2016)

The idea

Smaller loss under other winners, more weight
Losses p_ij, sum |f_jj|Weight w (F w = 0)Educational institutions (per 1000 residents)Educational institution…Kindergarten places: 0.6873Healthcare institutions (per 1000 residents): 0.6963Crime rate (per 1000 residents): 0.78732.17090.1277Kindergarten placesEducational institutions (per 1000 residents): 0.5176Healthcare institutions (per 1000 residents): 0.5204Recreational facilities (per 1000 residents): 0.5176Crime rate (per 1000 residents): 0.73812.29370.1924Healthcare institutions (per 1000 residents)Healthcare institution…Educational institutions (per 1000 residents): 0.0122Kindergarten places: 0.7940Recreational facilities (per 1000 residents): 0.0122Crime rate (per 1000 residents): 0.68501.50330.2993Recreational facilities (per 1000 residents)Recreational facilities…Kindergarten places: 0.8827Healthcare institutions (per 1000 residents): 0.8861Crime rate (per 1000 residents): 0.58132.35010.1180Crime rate (per 1000 residents)Crime rate (per 1000…Educational institutions (per 1000 residents): 0.6626Kindergarten places: 0.0447Healthcare institutions (per 1000 residents): 0.5956Recreational facilities (per 1000 residents): 0.66261.96560.2626

Each segment is the relative loss p_ij on a criterion when another criterion's best elderate is picked; their sum is |f_jj|, and F w = 0 gives the weights.

Show as table
CriterionEducational institutions (per 1000 residents)Kindergarten placesHealthcare institutions (per 1000 residents)Recreational facilities (per 1000 residents)Crime rate (per 1000 residents)Total loss |f_jj|Weight
Educational institutions (per 1000 residents)0.00000.68730.69630.00000.78732.17090.1277
Kindergarten places0.51760.00000.52040.51760.73812.29370.1924
Healthcare institutions (per 1000 residents)0.01220.79400.00000.01220.68501.50330.2993
Recreational facilities (per 1000 residents)0.00000.88270.88610.00000.58132.35010.1180
Crime rate (per 1000 residents)0.66260.04470.59560.66260.00001.96560.2626

Data: Vilnius neighbourhoods, social criteria (Zavadskas et al. 2017).

CILOS asks how much the other criteria lose when you pick the best alternative on one criterion.

When to use it

  • Data are positive.
  • You want to combine it with Entropy (IDOCRIW), Vilnius-school style.

Algorithm

  1. 1Turn cost criteria into benefit
  2. 2Sum normalization
  3. 3Matrix of column winners

    Row i of the n x n matrix A is the winning row on criterion i (ties are averaged for row-order invariance), so a_ii is the column maximum.

  4. 4Relative losses

    The loss on criterion j when the alternative that is best on criterion i is chosen.

  5. 5Loss matrix with column sums on the diagonal
  6. 6Weights from a homogeneous system

    The columns of F sum to zero, so F is singular by construction; a unique normalized solution exists when F has rank n - 1.

Used together with

Why these pairs
  • CILOS times Entropy gives IDOCRIW, which Zavadskas et al. (2017) then use with EDAS and other ranking methods.
  • CILOS and IDOCRIW weights paired with EDAS, SAW, TOPSIS and COPRAS on the Vilnius neighbourhood dataset: Zavadskas et al. (2017).
  • Compared with Entropy, IDOCRIW, CRITIC and D-CRITIC on one matrix by Krishnan et al. (2021).
  • Fuzzy extensions FCILOS and FIDOCRIW: Podvezko, Zavadskas & Podviezko (2020).

Reference example

Zavadskas, E.K.; Cavallaro, F.; Podvezko, V.; Ubarte, I.; Kaklauskas, A. (2017)Matches the printed digits
Details

Zavadskas, E.K.; Cavallaro, F.; Podvezko, V.; Ubarte, I.; Kaklauskas, A. (2017). MCDM Assessment of a Healthy and Safe Built Environment According to Sustainable Development Principles: A Practical Neighborhood Approach in Vilnius. Sustainability 9(5), 702. doi:10.3390/su9050702

Published example (Table 4, social and economic blocks; Tables 10 and 8, CILOS): twenty-one Vilnius neighbourhoods, five criteria per block. All published CILOS values are reproduced to 4 decimals, but the paper attaches them to the wrong criteria: in Table 10 the first two rows are swapped (and in Table 8 the rows follow [2, 4, 0, 1, 3]). Matched after undoing the swap.

Checked by this site's test suite.

Sources

Show 7 sources
  1. Zavadskas and Podvezko (2016). Integrated Determination of Objective CRIteria Weights in MCDM. International Journal of Information Technology & Decision Making 15(2), 267-283. doi:10.1142/S0219622016500036The idea goes back to Mirkin (1974), a Russian book on group choice; the algorithm follows the same authors' open-access paper Zavadskas et al. (2017), section 5.1.2.
  2. Zavadskas & Podvezko (2016), International Journal of Information Technology & Decision Making 15(2), 267-283 doi:10.1142/S0219622016500036
  3. Zavadskas, Cavallaro, Podvezko, Ubarte & Kaklauskas (2017), Sustainability 9(5), 702 doi:10.3390/su9050702
  4. Pala (2021), Kafkas Üniversitesi İİBF Dergisi 12(23), 263-294 doi:10.36543/kauiibfd.2021.013
  5. Krishnan, Kasim, Hamid & Ghazali (2021), Symmetry 13(6), 973 doi:10.3390/sym13060973
  6. Mirkin (1974), Problema gruppovogo vybora, Nauka, Moscow
  7. Podvezko, Zavadskas & Podviezko (2020), Economic Computation and Economic Cybernetics Studies and Research 54(2), 59-75 doi:10.24818/18423264/54.2.20.04