Keywords
watershed impact assessment, sustainable development, regression model, artificial neural network modeling, model comparison
Start Date
1-7-2012 12:00 AM
Abstract
Impact assessment of a micro-watershed management project has been carried out to evaluate sustainable livelihood security for local people especially, of developing countries. In general, the conventional approaches for impact assessment have been found to be time-consuming, expensive and the data generated through these studies are mostly unused in future. In order to overcome the deficiency of conventional impact assessment methods, the present study has targeted to develop suitable Regression and Artificial Neural Network (ANN) models using identified 144 randomly selected indicators data sets over nine years historical time periods, collected from a successful case study namely “Semri micro watershed, Sehore District, Madhya Pradesh, India”. Regression and ANN decision support system prediction models have been developed with eight most dominating parameters which have found most significant effect on livelihood security. The comparison study of these two models have indicated that, the statistical yield predicted through ANN models performed better than that predicted through regression models. The study has recommended the use of such models for improvement of similar degraded watershed for future reference.
Comparison of Regression and Artificial Neural Network Impact Assessment Models: A Case Study of Micro-Watershed Management in India.
Impact assessment of a micro-watershed management project has been carried out to evaluate sustainable livelihood security for local people especially, of developing countries. In general, the conventional approaches for impact assessment have been found to be time-consuming, expensive and the data generated through these studies are mostly unused in future. In order to overcome the deficiency of conventional impact assessment methods, the present study has targeted to develop suitable Regression and Artificial Neural Network (ANN) models using identified 144 randomly selected indicators data sets over nine years historical time periods, collected from a successful case study namely “Semri micro watershed, Sehore District, Madhya Pradesh, India”. Regression and ANN decision support system prediction models have been developed with eight most dominating parameters which have found most significant effect on livelihood security. The comparison study of these two models have indicated that, the statistical yield predicted through ANN models performed better than that predicted through regression models. The study has recommended the use of such models for improvement of similar degraded watershed for future reference.
