ارزيابي عملكرد شركت‌هاي آب و فاضلاب استان مازندران به كمك مدل تركيبي تحليل پوششي داده‌ها و شبكه عصبي مصنوعي

Performance Evaluation of Mazandaran Water and Wastewater by Data Envelopment Analysis and Artificial Neural Network

در اين تحقيق، به كمك تكنيك‌ تحليل پوششي داده‌هاي ورودي‌محور، شركت آب و فاضلاب استان مازندران ارزيابي مي‌شود. يكي از روش‌هاي ارزيابي كارايي واحدها روش تحليل پوششي داده‌ها[i] است. با توجه به كارايي محاسبه‌شده براي شانزده واحد تصميم‌گيرنده سال‌هاي 1389 و 1390، مشكل وجود چندين ناحيه كارآمد مشخص است. در مرحله اول براي رفع اين ضعف از تكنيك اندرسون و پيترسون[ii] استفاده شد. از آنجا كه تكنيك AP شامل حل يك مدل برنامه‌ريزي خطي به‌ازاي هر DMUs است، با افزايش بعد مسيله ارزيابي كارايي زمان‌بر خواهد شد. بنابراين، استفاده از شبكه‌‌اي عصبي به منزله رويكردي جايگزين مطرح مي‌شود. نتايج تحليلي كارايي‌هاي محاسبه‌شده DMUsها به كمك روش تركيبي Neuro-DEA حاكي از قدرت بالاي شبكه عصبي در تفكيك‌پذيري نواحي تصميم‌گيري است.

In this study, Mazandaran Water and Wastewater Company’ performance is evaluated by using an input-oriented data envelopment analysis. As a principle, the performance of each organizational unit or organization should be measured as far as possible and what cannot be evaluated cannot be well governed. One method of evaluating the performance of units is data envelopment analysis method. One of the main problems of using data envelopment analysis is its low-resolution which it is due to the low number of decision making units to compare with the number of inputs and outputs. Given to the calculated efficiency by the DEA model (CCR input-oriented) for 16 decision making unit for years 1389 and 1390 there is the problem of existence of several efficient areas, which in the first step was used from Anderson and Peterson (AP) technique to cover this weaknesses. Since the AP technique involves solving a linear programming model for each of the DMUs. Therefore, by increasing the dimension of issue, efficiency assessment will be time consuming process. So the idea of using a neural network with efficiency data of data envelopment analysis is proposed as an alternative approach. Analytical results of calculated efficiencies of DMUs by the combination method of Neuro-DEA indicate the high power of neural network in resolution of decision-making areas in terms of efficiency.