Abstract:
Today, finding material, good, and semi-finished parts suppliers in a competitive atmosphere is significantly easier, therefore there are many options to select from for commercial partners. On the other hand, purchase responsibility as a management decision has challenged company managers due to its complexity and variety of evaluation measures. The present study provides an integrated model using data envelopment analysis (DEA) techniques to develop decision support systems in an auto part supplier, and methods of data mining like neural networks in order to evaluate suppliers. Efficiency scores of each supplier are obtained by solving model of “data envelopment analysis”, and then this model, using educational data, trains artificial neural networks to predict and rank suppliers. Results of the selected model provide a complete ranking and an appropriate grouping with an acceptable level of prediction accuracy respectively to evaluation decision making and selection of suppliers. The main objective of this study is to evaluate performance of logistic service suppliers, and also seeks to find an answer about how to use methods of data envelopment analysis and data mining in evaluation of suppliers. Time percentage measure of “timely delivery” of parts has the maximum efficiency and the measure of “ability to reduce price” has the minimum efficiency. In the other hand, once this model is used, there is no need to modeling and resolve data envelopment analysis models with high computational volume. Indeed, application of neural networks has been able to eliminate defect of prediction disability in data envelopment analysis.
Machine summary:
"The present study provides an integrated model using data envelopment analysis (DEA) techniques to develop decision support systems in an auto part supplier, and methods of data mining like neural networks in order to evaluate suppliers.
Evaluating vendors in supply chain is one of the most important processes of decision-making that includes not only vendors, but also other decisions like order quantities of each vendor [3] Method of data envelopment analysis (DEA) is one of the tools to evaluate suppliers and relative comparing of them.
Data mining can be suitable for analyzing different DEA models Table 1: different measures of supplier selection [6] [7] Data mining approaches such as neural networks and expert systems are used to predict performance of a new manufacturer.
In their study, integrated method of DEA and MDE is used [4] Developed a decision-maker support system using DEA to monitor efficiency based on organization`s key performance in order to evaluate and manage relative performance of the organization [5] In the present study, our goal is to integrate DEA technique with data mining methods to eliminate restrictions such as disability to predict, impossibility of comparing heterogeneous units, and disability of DEA basic models to specify measurement error; therefore, we use neural networks method.
Table 8: average rate of changes relative to each of input variables The results of sensitivity analysis show that variable of "time percentage of timely delivery DELT" is a variable with greatest effect on suppliers` supply efficiency, and this is reasonable because failure to timely delivery of parts to an automaker company is practically followed by production line suspension and incurring high expenses to management."