Preprocessing of Missing Values Using Robust Association Rules

1 I n t r o d u c t i o n The missing values problem is an old one for analysis tasks[8] [11]. The waste of da t a which can result f rom casewise deletion of missing values, obliges to propose alternatives approaches. A current one is to t ry to determine these values [9]. However, techniques to guess the missing values mus t be efficient, otherwise the complet ion introduces noise. Wi th the emergence of K D D for industrial databases, where missing values are inevitable, this problem has become a priori ty task [6] also requiring declarat ivi ty and interactivi ty during t rea tments . At the present t ime, t rea tments are often specific and internal to the methods , and do not offer such qualities. Consequently the missing values problem is still a challenging task of the KDD research agenda [6]. We have proposed in [14] the R A R algor i thm to correct the weakness of usual association rules algorithms[2] in mining databases with mult iple missing values. The efficiency of this a lgor i thm to extract quickly all the associations contained in such a database, allows to use it for the missing values problem. T h a t is what 1 Knowledge Discovery in Databases 2 Robust Association Rules