Using an Efficient Optimal Classifier for Soil Classification in Spatial Data Mining Over Big Data

Abstract

This article proposes an effectual process for soil classification. The input data of the proposed procedure is the Harmonized World Soil Database. Preprocessing aids to generate enhanced representation and will use minimum time. Then, the MapReduce framework divides the input dataset into a complimentary portion that is held by the map task. In the map task, principal component analysis is used to reduce the data and the outputs of the maps are then contributed to reduce the tasks. Lastly, the proposed process is employed to categorize the soil kind by means of an optimal neural network (NN) classifier. Here, the conventional NN is customized using the optimization procedure. In an NN, the weights are optimized using the grey wolf optimization (GWO) algorithm. Derived from the classifier, we categorize the soil category. The performance of the proposed procedure is assessed by means of sensitivity, specificity, accuracy, precision, recall, and F-measure. The analysis results illustrate that the recommended artificial NN-GWO process has an accuracy of 90.46%, but the conventional NN and k-nearest neighbor classifiers have an accuracy value of 75.3846% and 75.38%, respectively, which is the least value compared to the proposed procedure. The execution is made by Java within the MapReduce framework using Hadoop.

Keywords

• mapreduce framework
• principal component analysis
• neural network
• grey wolf optimization
• accuracy
• precision
• recall
• f-measure