|
Useful Links |
|
|
Article Details :: |
|
|
| Article Name : | | | ENSEMBLE APPROACH FOR RULE EXTRACTION IN DATA MINING. | | Author Name : | | | HITESH NINAMA | | Publisher : | | | Ashok Yakkaldevi | | Article Series No. : | | | GRT-2636 | | Article URL : | |  | Author Profile View PDF In browser | | Abstract : | | | A major drawback with neural networks is that the models produced are opaque; i.e. they do not permit human inspection or understanding. A decision tree model, on the other hand, is regarded as comprehensible since it is transparent, making it possible for a human to follow and understand the logic behind a prediction. Although accuracy is the prioritized criterion for predictive modeling, the comprehensibility of the model is often very important. A comprehensible model makes it possible for the user to understand not only the model itself but also why individual predictions are made. Traditionally, most research papers focus on high accuracy, although the comprehensibility criterion is often emphasized by business representatives. Clearly, comprehensibility is very important for data mining technique. Since techniques producing opaque models normally will obtain highest accuracy, it seems inevitable that the choice of technique is a direct trade-off between accuracy and comprehensibility. With this trade-off in mind, several researchers have tried to bridge the gap by introducing techniques for transforming opaque models into transparent models, keeping an acceptable accuracy. Most significant are the many attempts to extract rules from trained neural networks. And this technique of transforming opaque model into transparent model is called as Rule Extraction (RE). Within the machine learning research community it is, however, also well known that it is possible to obtain even higher accuracy, by combining several individual models into ensembles. The overall goal when creating an ensemble is to combine models that are highly accurate, but differ in their predictions. The common ensemble techniques are probably bagging, boosting and stacking, all of which can be applied to different types of models and perform both regression and classification. Most importantly; bagging, boosting and stacking will, almost always, increase predictive performance over a single model. The proposed approach converts the opaque model into transparent model and at the same time maintains acceptable level of accuracy. Our idea is to first produce an opaque model by applying a data mining algorithm neural network to produce accurate model that optimizes the accuracy and then extract the rules from this opaque model by applying rule extraction algorithm to convert the opaque model into transparent model. Further we apply the same algorithm multiple times i.e. to devise ensemble method for rule extraction that leverages the power of multiple methods to achieve greater comprehensibility than any individual approach. Experiments are conducted to show that the proposed work reduces the accuracy vs comprehensibility tradeoff. | | Keywords : | | - Rule Extraction, Data Mining, Decision Tree, Ensemble approach, Bagging, Boosting, Neural Network, Accuracy, Comprehensibility.,
|
|
|