Applications of Data Mining in Weather Forecasting Using Frequent Pattern Growth Algorithm

Rainfall forecasting or Weather forecasting has been one of the most challenging problems around the world because it consists of multidimensional and nonlinear data such as in the field of agriculture to determine initial growing season. Recently, climate change causes much trouble in rainfall forecasting. Our Project describes five data mining algorithms namely neural network (NN), random forest, classification and regression tree (CRT), support vector machine (SVM) and k-nearest neighbour. Generally these algorithms are used for the prediction. Generally these algorithms used for prediction. In this project we use five years previous data from Jan 2010-Jan 2014 for Nagpur station. On available datasets we apply the Frequent Pattern Growth Algorithm for deleting the inappropriate data. Generally temperature, humidity, wind speed are mainly responsible for the rainfall prediction. On the percentage of these parameters we predict there is a rainfall or not. Based on experiment result, it can be concluded that the combination of GA and FP growth algorithm weather data can gives prediction with higher than 90 % accuracy with several population size and crossover probability.