论文标题
数据预处理和评估几种数据挖掘方法的性能，以预测灌溉水的需求
Data Pre-Processing and Evaluating the Performance of Several Data Mining Methods for Predicting Irrigation Water Requirement
论文作者
论文摘要
最近的干旱和人口增长正在种植对使用有限水资源的前所未有的需求。灌溉农业是淡水的主要消费者之一。由于水管理习惯不佳，灌溉农业中的大量水浪费了。为了改善灌溉区域的水管理，需要估算未来水需求的模型。开发预测灌溉水需求的模型可以改善水管理实践并最大程度地提高水生产率。数据挖掘可有效地用于构建此类模型。 在这项研究中，我们准备一个数据集，其中包含有关预测灌溉水需求的合适属性的信息。数据是从三种不同来源获得的，即气象数据，遥感图像和水输送声明。为了使准备好的数据集对需求预测和模式提取有用，我们根据灌溉和数据挖掘知识的组合使用新颖的方法预处理数据集。然后，我们应用和比较不同数据挖掘方法的有效性，即决策树（DT），人工神经网络（ANN），系统开发的森林（SYSFOR），用于多棵树，支持向量机（SVM），逻辑回归和传统的蒸发液刺激（ETC）方法（ETC）方法，并评估这些模型的性能以预测这些模型以预测灌溉水需求。我们的实验结果表明，数据预处理和不同分类器的有效性的有用性。在我们使用的六种方法中，SYSFOR以97.5％的精度产生最佳预测，然后是96％的决策树和ANN，分别通过将预测与实际用水密切匹配，分别为95％。因此，我们建议使用SYSFOR和DT模型进行灌溉水的预测。
Recent drought and population growth are planting unprecedented demand for the use of available limited water resources. Irrigated agriculture is one of the major consumers of freshwater. A large amount of water in irrigated agriculture is wasted due to poor water management practices. To improve water management in irrigated areas, models for estimation of future water requirements are needed. Developing a model for forecasting irrigation water demand can improve water management practices and maximise water productivity. Data mining can be used effectively to build such models. In this study, we prepare a dataset containing information on suitable attributes for forecasting irrigation water demand. The data is obtained from three different sources namely meteorological data, remote sensing images and water delivery statements. In order to make the prepared dataset useful for demand forecasting and pattern extraction, we pre-process the dataset using a novel approach based on a combination of irrigation and data mining knowledge. We then apply and compare the effectiveness of different data mining methods namely decision tree (DT), artificial neural networks (ANNs), systematically developed forest (SysFor) for multiple trees, support vector machine (SVM), logistic regression, and the traditional Evapotranspiration (ETc) methods and evaluate the performance of these models to predict irrigation water demand. Our experimental results indicate the usefulness of data pre-processing and the effectiveness of different classifiers. Among the six methods we used, SysFor produces the best prediction with 97.5% accuracy followed by a decision tree with 96% and ANN with 95% respectively by closely matching the predictions with actual water usage. Therefore, we recommend using SysFor and DT models for irrigation water demand forecasting.
