Limits...
Evaluation models for soil nutrient based on support vector machine and artificial neural networks.

Li H, Leng W, Zhou Y, Chen F, Xiu Z, Yang D - ScientificWorldJournal (2014)

Bottom Line: Traditional evaluation approaches of soil nutrient are quite hard to operate, making great difficulties in practical applications.We took the content of organic matter, total nitrogen, alkali-hydrolysable nitrogen, rapidly available phosphorus, and rapidly available potassium as independent variables, while the evaluation level of soil nutrient content was taken as dependent variable.In practical applications, both SVM and GRNN models can be used for determining the levels of soil nutrient.

View Article: PubMed Central - PubMed

Affiliation: College of Chemistry, Sichuan University, Chengdu, Sichuan 610064, China ; Key Laboratory of Marine Bio-Resources Restoration and Habitat Reparation in Liaoning Province, Dalian Ocean University, Dalian 116023, China.

ABSTRACT
Soil nutrient is an important aspect that contributes to the soil fertility and environmental effects. Traditional evaluation approaches of soil nutrient are quite hard to operate, making great difficulties in practical applications. In this paper, we present a series of comprehensive evaluation models for soil nutrient by using support vector machine (SVM), multiple linear regression (MLR), and artificial neural networks (ANNs), respectively. We took the content of organic matter, total nitrogen, alkali-hydrolysable nitrogen, rapidly available phosphorus, and rapidly available potassium as independent variables, while the evaluation level of soil nutrient content was taken as dependent variable. Results show that the average prediction accuracies of SVM models are 77.87% and 83.00%, respectively, while the general regression neural network (GRNN) model's average prediction accuracy is 92.86%, indicating that SVM and GRNN models can be used effectively to assess the levels of soil nutrient with suitable dependent variables. In practical applications, both SVM and GRNN models can be used for determining the levels of soil nutrient.

Show MeSH

Related in: MedlinePlus

Results of SVM model: (a) different results calculated from diverse normalization conditions in model 1; (b) different results calculated from diverse normalization conditions in model 2.
© Copyright Policy - open-access
Related In: Results  -  Collection


getmorefigures.php?uid=PMC4273551&req=5

fig4: Results of SVM model: (a) different results calculated from diverse normalization conditions in model 1; (b) different results calculated from diverse normalization conditions in model 2.

Mentions: Figure 4 indicates the fluctuations of the prediction accuracy calculated in diverse normalization condition. We computed each model for 1000 times, axis Y represents the value of the prediction accuracy of each count, whereas axis X just stands for the time. That is to say, the first point of the picture is the prediction accuracy calculated for the first time, and the second point represents the second results. The last point is the value of the prediction calculation in the last calculation. The fluctuation of the prediction accuracy illustrates the steady level of the model. Color pink represents the data that is normalized in (0, 1), color green stands for the data that is pretreated under the normalization condition (−1, 1), and color violet is the result calculated from the data without normalization.


Evaluation models for soil nutrient based on support vector machine and artificial neural networks.

Li H, Leng W, Zhou Y, Chen F, Xiu Z, Yang D - ScientificWorldJournal (2014)

Results of SVM model: (a) different results calculated from diverse normalization conditions in model 1; (b) different results calculated from diverse normalization conditions in model 2.
© Copyright Policy - open-access
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC4273551&req=5

fig4: Results of SVM model: (a) different results calculated from diverse normalization conditions in model 1; (b) different results calculated from diverse normalization conditions in model 2.
Mentions: Figure 4 indicates the fluctuations of the prediction accuracy calculated in diverse normalization condition. We computed each model for 1000 times, axis Y represents the value of the prediction accuracy of each count, whereas axis X just stands for the time. That is to say, the first point of the picture is the prediction accuracy calculated for the first time, and the second point represents the second results. The last point is the value of the prediction calculation in the last calculation. The fluctuation of the prediction accuracy illustrates the steady level of the model. Color pink represents the data that is normalized in (0, 1), color green stands for the data that is pretreated under the normalization condition (−1, 1), and color violet is the result calculated from the data without normalization.

Bottom Line: Traditional evaluation approaches of soil nutrient are quite hard to operate, making great difficulties in practical applications.We took the content of organic matter, total nitrogen, alkali-hydrolysable nitrogen, rapidly available phosphorus, and rapidly available potassium as independent variables, while the evaluation level of soil nutrient content was taken as dependent variable.In practical applications, both SVM and GRNN models can be used for determining the levels of soil nutrient.

View Article: PubMed Central - PubMed

Affiliation: College of Chemistry, Sichuan University, Chengdu, Sichuan 610064, China ; Key Laboratory of Marine Bio-Resources Restoration and Habitat Reparation in Liaoning Province, Dalian Ocean University, Dalian 116023, China.

ABSTRACT
Soil nutrient is an important aspect that contributes to the soil fertility and environmental effects. Traditional evaluation approaches of soil nutrient are quite hard to operate, making great difficulties in practical applications. In this paper, we present a series of comprehensive evaluation models for soil nutrient by using support vector machine (SVM), multiple linear regression (MLR), and artificial neural networks (ANNs), respectively. We took the content of organic matter, total nitrogen, alkali-hydrolysable nitrogen, rapidly available phosphorus, and rapidly available potassium as independent variables, while the evaluation level of soil nutrient content was taken as dependent variable. Results show that the average prediction accuracies of SVM models are 77.87% and 83.00%, respectively, while the general regression neural network (GRNN) model's average prediction accuracy is 92.86%, indicating that SVM and GRNN models can be used effectively to assess the levels of soil nutrient with suitable dependent variables. In practical applications, both SVM and GRNN models can be used for determining the levels of soil nutrient.

Show MeSH
Related in: MedlinePlus