Provided by: International Journal of Computer Applications
Topic: Big Data
Date Added: Nov 2012
Today, several tools are available for solving data mining problems, both in open source and commercial category. For solving classification problems these tools provide variety of strategies such as decision tree, neural networks, lazy classifiers etc. For each strategy, the tools allow the user to select specific values for large number parameters for e.g. in case of a neural network classifier, user needs to provide values for parameters such as epochs, learning rate, momentum etc. Although default setting for such parameters is provided by tools, it is often found that the classifier performance (accuracy) may be enhanced by making series of experiments with different values for these parameters.