DC (Drought Classifier): Forecasting and Classification of Drought Using Association Rules

Normally droughts are viewed as the nature disasters which show heavy economic impact in the affected regions. Indemnifying the information about the pattern, area, severity and timing of droughts effect, can be used for operational planning and decision making. In this work, combination of Artificial Neural Network (ANN) coupled with Fuzzy C-means and association rule mining are used to develop a model to identify the severity of drought by forecasting the climate conditions for upcoming season. A suitable Feed Forward Neural network (FFNN) is developed with forward selection to forecast the rainfalls for future years with the input dataset of several archived data. Later fuzzy c-means (FCM) clustering is used for partitioning the forecasted data in three groups like low, medium and high rainfall. Finally association rules are used to find associations among data belonging to the climate information using proposed rule based model. The low rain data group generated by FCM is used for classifying the drought effect from the predicted results.