Decision stream
Decision stream is a statistic-based supervised learning technique.[1] It generates a deep directed acyclic graph of decision rules to solve classification and regression tasks. This method avoids the problem of data exhaustion in terminal nodes. It merges the leaves from the same/different levels of predictive model (Fig. 1).
Decision stream provides:
- High accuracy: precise splitting of data with test statistics.
- Decrease of overfitting: partition of data into statistically representative groups.
- Reduction of complexity on every level of predictive model.
- Self-regulated depth of predictive model.
This method builds a predictive model with a high degree of connectivity. It merges statistically indistinguishable nodes at each iteration. With the same quantity of nodes, it provides a higher depth than a decision tree (Fig. 2). Decision stream splits and merges the data multiple times with different features.
References
- ↑ 1.0 1.1 Ignatov, D.Yu.; Ignatov, A.D. (2017). "Decision Stream: Cultivating Deep Decision Trees". IEEE International Conference on Tools with Artificial Intelligence: 905–912. arXiv:1704.07657. doi:10.1109/ICTAI.2017.00140.
This article "Decision stream" is from Simple English Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Decision stream.
