Generating rules from examples
Bijan Arbab, Donald Michie · 1985
Automatic design-tree generators have been used in production of expert systems. A number of experiments indicate the need for more linear (and more understandable) yet efficient decision trees. An algorithm is implemented for constructing decision trees optimized with respect to linearity. It also improves on a previous linearizing algorithm (AOCDL) with repect to execution efficiency.