Comparing multivariate regression and artificial neural network to predict barley production from soil characteristics in northern Iran uri icon

abstract

  • In this study artificial neural network (ANN) models were designed to predict thebiomass and grain yield of barley from soil properties; and the performance ofANN models was compared with earlier tested statistical models based onmultivariate regression. Barley yield data and surface soil samples (0?30 cmdepth) were collected from 1 m2 plots at 112 selected points in the arid regionof northern Iran. ANN yield models gave higher coefficient of determination andlower root mean square error compared to the multivariate regression, indicatingthat ANN is a more powerful tool than multivariate regression. Sensitivityanalysis showed that soil electrical conductivity, sodium absorption ratio, pH,total nitrogen, available phosphorus, and organic matter consistently influencedbarley biomass and grain yield. A comparison of the two methods to identify themost important factors indicated that while in the ANN analysis, soil organicmatter (SOM) was included among the most important factors; SOM wasexcluded from the most important factors in the multivariate analysis. Thissignificant discrepancy between the two methods was apparently a consequenceof the non-linear relationships of SOM with other soil properties. Overall, ourresults indicated that the ANN models could explain 93 and 89% of the totalvariability in barley biomass and grain yield, respectively. The performance of theANN models as compared to multivariate regression has better chance forpredicting yield, especially when complex non-linear relationships exist amongthe factors. We suggest that for further potential improvement in predictingthe barley yield, factors other than the soil properties considered such as soilmicronutrient status and soil and crop management practices followed during thegrowing season, need to be included in the models

publication date

  • 2011
  • 2011