skcriteria.agg.ocra module
Implementation of OCRA (Operational Competitiveness Rating) method for general MCDM purposes.
- skcriteria.agg.ocra.ocra_performance(matrix, objectives, weights)[source]
Compute the overall performance of each alternative.
- class skcriteria.agg.ocra.OCRA[source]
Bases:
SKCDecisionMakerABCOCRA (Operational Competitiveness Rating) method.
OCRA was initially intended (Parkan, 1994) to maximize the efficiency of a Production Unit (PU), seen as a set of activities that consume resources (inputs) and generate rewards (outputs), thus leading to a higher operational competitiveness. OCRA is thought of as an improvement of Data Envelopment Analysis (DEA): more efficient, robust, and properly sensitive to changes in inputs or outputs.
In a general-purpose sense, PUs are the alternatives to be compared, and a quantity and value of each type of input/output is instead given as a criteria value and corresponding weight. Inputs are non-beneficial criteria that should be minimized, and Outputs are beneficial criteria that should be maximized; thus, the entire Decision Matrix is used.
The performance of beneficial and non-beneficial criteria is computed separately and aggregated for each alternative, and then I/O criteria are summed to yield a final performance ranking. The Min value of each criteria is used in both cases (rather that Max - Min), following the implementation from Işık, A. T. (2016) and Madić, M. (2015). This means values are not scaled as 0-1 (possible future extension); however, they are floored to the Min value twice (first separately, then overall), such that the worst performance is always zero.
\[I_i = \sum_{j=1}^{g} w_j \frac{\max{x_{ij}} - x_{ij}}{\min{x_{ij}}}\ O_i = \sum_{j=g+1}^{n} w_j \frac{x_{ij} - \min{x_{ij}}}{\min{x_{ij}}}\ for\ i = 1,2,3,...,m\]with: \(j = 1, 2, ..., g\) for the objectives to be minimized, \(j = g + 1, g + 2, ..., n\) for the objectives to be maximized. \(w_j\) is the relative importance (weight) of each criteria.
References