Partially labelled data has some data items with labels, but soem wth none. Supervised learning depends on labelled training data, but it may be difficult to obtain fully labelled data, either becasue of availability of data or because {[expert labelling}} is expensive. However, some algorithms can be modified to work with partially labelled data; for example the k-means algorithm can use a combination of labelled and unlabelled data.
Used in Chap. 9: pages 139, 140
Also known as partial labelling