Arbitrary Decision Tree for Weather Prediction

Long before technology was developed, folks had to trust observations, patterns and their expertise to predict the weather. Here we present novel approach for predicting weather using decision tree, it is based on concepts and techniques from data mining and prediction systems. The suitability of this approach for prediction and its advantages compared with different techniques are considered here. This paper highlights on development of weather prediction model using decision tree and implementation of it. Concepts of entropy is used to find the homogeneity of classes present in the dataset, information gain used for finding the threshold values for the nodes and finally the Hoeffding bounds [1] of inequality to choose the minimum number of split examples from the dataset. It selects attributes randomly and constructs a tree which will be efficient and will improve the accuracy of classifiers.