Performance Metrics for Model Fusion in Twitter Data Drifts

摘要: Ensemble approaches have revealed remarkable abilities to tackle different learning challenges, namely in dynamic scenarios with concept drift, e.g. social networks, as Twitter. Several efforts been engaged defining strategies combine the models that constitute an ensemble. In this work, we investigate effect of using metrics for combining ensembles’ models, specifically performance-based metrics. We propose five performance metrics, having mind may take advantage diversity classifiers, their individual takes a leading role contribution Experimental results on Twitter dataset, artificially timestamped, suggest ensemble can introduce relevant improvements overall performance.