Propositionalisation of Continuous Attributes beyond Simple Aggregation

Flexible propositionalization of continuous attributes in relational data mining

“…It is shown in El Jelali et al (2012) that no single propositionalization and attribute-value learner wins on all datasets. Moreover, we discussed the importance of finding an optimal combination on the training dataset in Section 1.…”

“…In addition, we propose an effective wrapping algorithm to select the best combination of propositionalization and classification methods which adds more flexibility in relational data mining for online/incremental/data stream environment. This paper includes substantively new and different contributions beyond the preliminary conference version (El Jelali, Braud, & Lachiche, 2012) including detailed presentations of the approaches with revised and optimized algorithms, enhanced motivations with real-life applications, extended experiments with new classifiers and more real-life benchmarks, introduction and evaluation of a wrapping algorithm to identify the best combination of propositionalization and classification methods from training data.…”

Reframing in context: A systematic approach for model reuse in machine learning

“…As another example, the input data can be transformed into a normalised or a discretised version [17], using the data distribution. Note that in this case we apply the transformation to the input features both during training and during testing.…”