Self-Organizing Map and Multi-Layer Perceptron Neural Network Based Data Mining to Envisage Agriculture Cultivation

Dr P Thangaraj,E.T Venkatesh

doi:10.3844/jcssp.2008.494.502

Abstract

Study on characteristics of soil, to determine the types of crops suitable for cultivation in a particular region can increase the yield to greater extent, which minimizes the expenditures involved in irrigation and application of fertilizers. With the tested techniques available for calibrating the quality of soil and the crops suitable for cultivation in it, it is possible to determine the exact crop, irrigation patterns and even the cycle and quantity of fertilizer application. This paper dealt with the application of SOM based clustering and Artificial Intelligence techniques, to analyze the patterns of soils distributed across huge geographical area and identify the suitable types of crops for the particular soil. Estimation of exact crop(s) suitable for a particular region can help stave off redundant maintenance and the inherent expenditures that would occur due to over irrigation and over usage of fertilizers, to fulfill the natural deficiencies. Our Focus is to improve the optimal utilization of innate characteristics in a soil through cultivation of appropriate crops, which will increase the volume and quality of yield, in particular for a developing country like India, where the huge majority of the population depends primarily on agriculture for livelihood.

Highlights

Information packaged in sizeable databases about all kinds of characteristics and facets are accessible these days
With selforganizing map (SOM) as the basis we have proposed an approach for clustering data items
The application of Self Organizing Maps for clustering of data and Multi-Layered Perceptron Neural Network or Multilayer Feed-Forward Neural Network 3. has helped achieving near precise estimation of suitability of soil characteristics and the choice of crops planned for cultivation based on its nutrient requirements

Summary

Introduction

Information packaged in sizeable databases about all kinds of characteristics and facets are accessible these days. Extracting new and interesting knowledge through interpreting the collected data is cumbersome since more data potentially contains more information. To analyze these databases, immense efforts are being deployed. Understanding the hidden are implicit relationship between attributes in these Knowledge Discovery Databases (KDD) requires data mining techniques, as it has been proven as an effective tool. Data mining in nutshell is, pattern finding. Data mining is the process of discovering previously unknown and potentially interesting patterns in large datasets [1]. Finding patterns and elucidating those patterns clearly are the two primary activities of data mining. Facilitating to make profitable predictions through offering insights into the data and providing proper explanations about the data is the mark of efficient data mining tool [2]

Methods

Results

Conclusion