| 摘要 |
<p>The invention concerns a method for discretizing an attribute of a database containing a population of individuals, said attribute, called source attribute, capable of assuming several modalities. Said method is characterised in that, in a first step, it consists in assembling said modalities of the source attribute into elementary groups, and on the basis of the contingency table of the source attribute and of a target attribute, in determining, in a second step, among a set of pairs of elementary groups, the pair of groups whereof the merging most highly decreases independence probability of the source attribute and of the target attribute, and in merging in a third step the pair of elementary groups thus determined, said second and third steps being repeated as long as there is a pair of elementary groups enabling to decrease said independence probability.</p> |
