Databases, when used to collect data in the dynamic world, change over time. As a consequence, different parts of a database that come into the database at different times may contain activities with different characteristics. In other words, the data in the database are “non-uniform”. This non-uniformity of the database gives rise to the idea of a Divide and Conquer strategy that divides a very large and non-uniform database into component parts for better mining. Experiments conducted show that the method is effective under various situations.