Design an Intelligent Miner for Huge Medical Databases

In this work the suggested system is a Miner of obtain accuracy and comprehensible classified association rules (MOACCR). In this programming system we aim to extraction knowledge base in the form of classified association rules.Our suggested system consists of three process stages each one processed to handle one of the main challenges in KDD.In first process stage we build a novel tool called Developed Random Forest Local Least Square(DRFLLS) to estimate the optimal missing values in the databases having missing values.In the second process stage we proposed adboosting algorithm as we proposed adboosting algorithm as one of the ensemble classification algorithm to find the class for each record.In the third process stage we try to satisfy the goal of reducing the dimensions. We suggest using the correlation structure among the predicator variables in order to reduce in three main dimensions (features samples and value of features). We suggest new algorithm called Frequency Pattern-Knowledge Construction (FP-KC).The suggest algorithm (FP-KC) generation collection of classified association rules. The result is a collection of classified association rules.