A hybrid clustering and classification technique for soil data mining

Abstract

Predictive soil modelling using geostatistical methods is a research concept in modern soil science and soil geography for the last two decades. One of the reasons for this lack of soil spatial data is that conventional soil survey methods are relatively slow, qualitative and expensive. Spatial data sets covering large areas like digital geomorphographical maps, geological, land use, and climate data are available and these geo-datasets contain information about soil formation and resulting hydrologic variables etc which are needed to extract relevant soil information. In this paper we present an efficient hybrid model that was achieved by first clustering the data and then classifying it, and using the spatial conceptual information extracted from the environmental variables. This paper assists in assessment of the status of food production associated with land degradation and estimate indicators of soil nutrient mining by a country and region. The findings and conclusions of this paper result from the monitoring of the nutrient mining of agricultural lands in a country which have a direct implication on policy development. We propose a framework where soil is classified into different types, then a future work could be to predict soil fertility, based on which you can decide upon the fertilisers and suitable crops that could be cultivated with expertise.