The comparative analysis of various classification models on land evaluation
2009
Many methods of data mining model were widely applied for land evaluation, and they show different characteristics of
the application for land evaluation. In order to analyze different classification model effect for land evaluation, this paper
took land in Longchuan County as a case study, established three models using decision tree, back propagation neural
network (BP) and logistic regression on land evaluation. The result of study shows that the accuracy of three models
changes remarkably according to 6 groups of training samples. The accuracy of the decision tree and BP model can reach
high level in support of 4000 training samples, but decision tree model is superior to BP model at intelligibility of model
and consuming-time aspects. The overall performance of Logistic regression model is worse than other models at the
massive samples. Moreover, three model have different the characteristic of error distribution by means of confusion
matrix. The error of decision tree distributes evenly, and the error distribution of BP has opposite result of Logistic
regression. Results indicate that the model of decision tree is the best model for evaluating Longchun County land at
comprehensive thought, and it has a good effect on application.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
0
References
2
Citations
NaN
KQI