A classification framework for exploiting sparse multi-variate temporal features with application to adverse drug event detection in medical records

Table 2 For each ADE dataset, the number of features included in the learning process with different sparsity requirements

In particular, each column corresponds to the maximum percentage of empty time series which is tolerated for a dataset. When τ_sp=1.0, all the available features are taken into account, regardless of the percentage of empty sequences; the only requirement for a feature to be selected in the latter case is that it contains at least one non-empty sequence

ISSN: 1472-6947