Abstract: To induce a fuzzy decision tree from a continuous decision table,a method based on loose lower approximation in fuzzy rough set theory is proposed.First,a fuzzy decision table is generated by clustering fuzzy similarity relations.Second,based on loose lower approximation,a measure of the importance of a condition attribute is introduced.Finally,using this measure as a criterion for selecting the expanding attribute,a fuzzy decision tree induction algorithm is proposed and an illustrative example is provided.

Key words: continuous decision table, loose lower approximation, fuzzy decision tree