Efficient Decision Tree Construction for Classifying Numerical Data

Many organizations today have very large databases which grow at very fast rate. Efficient mining techniques are necessary to extract useful information from them. Performing classification on data streams with traditional classification algorithm based on decision tree has relatively poor efficiency in time and space. We made an attempt to create a model which will improve accuracy of classifier. The efficient decision tree construction algorithm uses Hoeffding bound along with information gain to select split point. Since it selects attributes randomly construction of tree is efficient hence it improves accuracy of classifier. Time required for classification is also improved for moderate datasets.