Pattern Recognition with Partly Missing Data

摘要: An experimental comparison of several simple inexpensive ways doing pattern recognition when some data elements are missing (blank) is presented. Pattern methods usually designed to deal with perfect data, but in the real world often due error, equipment failure, change plans, etc. Six dealing blanks tested on five sets. Blanks were inserted at random locations into A version K-nearest neighbor technique was used classify and evaluate six methods. Two found be consistently poor. Four generally good. Suggestions given for choosing best method a particular application.