With the recent boom in affordable sensor technologies, there is immense interest in the agricultural community in using sensor data to make smart decisions about farms. An important application in this field is the creation of management zones for fine control of irrigation schedule. Data from soil electrical conductivity sensors and GPS are used with clustering techniques to delineate soil types on the farm and create management zones.
In this project, we developed a general purpose, scalable web service that applies very well to the problem of creating management zones. The system performs multi-dimensional clustering and provides recommendations along with helpful visualizations. We developed a modified version of the K-Means algorithm that uses Mahalanobis distance to provide better clustering for correlated datasets and Bayesian Information Criterion (BIC) to determine the number of clusters that best fit the data.