Monitoring weather based meteorological data: Clustering approach for analysis

With the advancement in data mining and its applications, data mining is widely used to make smarter decisions in farming. Agricultural meteorology is the science that applies knowledge in weather data relating to atmospheric environment observed by instruments on the ground and by remote sensing. Most of the data need to be processed for generating various decisions such as cropping and scheduling of irrigation. This paper describes a data mining study of agricultural meteorological patterns collected from meteorological centre of Bengaluru district. We use K Means and Hierarchical clustering techniques to extract patterns like minimum (15 to 17°C) and maximum (28 to 29°C) air temperature, relative humidity (79 to 96%) in the interim of morning hours and (42 to 50%) in the interim of noon hours, rainfall (0 mm) and pan evaporation (5.22 to 7.2 mm) which gives great significance to predict probable result. The obtained results play a crucial role in the decision making for sustainable agriculture. Along with this we also compared these algorithms by applying Connectivity, Silhouette width and Dunn index formula which measures internal validation of clustering techniques. The results indicate that Hierarchical technique performs better than K Means in terms of Connectivity (10.1671), Silhouette width (0.4084) and Dunn index (0.4619).