Re[3]: Top Ten Errors In Data Wareousing

>A great deal of analysis work does indeed go into the warehouse as well as >in coding the 'strip' programs or 'fetch' programs (depending on whether >you are pushing or pulling). The guesstimate is that 80% of the effort >goes here. One of our criteria for inclusion of data into the warehouse >is that it be somewhat stable. Even so you have to plan to handle some >corrections. Our plan is to allow for these corrections and to monitor >them. As the number of corrections increases a corresponding 'confidence >in the data' value decreases for that data. That 'confidence' number is >kept in plain view where the user can easily see it. As the data from this application (and the need that drove the warehouse idea) are visible at a very high political level, confidence in the data was REQUIRED to be very high. As such, we could not get anyone at the management level to determine when the data would be considered "stable" and cease the updates. Because of the nasty nature of the politics in place, we decided to wait it out. As a FINAL result of the nasty politics, I moved to another contract. (long sob story deleted). I was once told by a datacomm instructor that the OSI 7-layer model is being revised to include 3 upper levels, FINANCIAL, POLITICAL and RELIGIOUS (8-10). After working in govt. contracting for about 3 years, I am beginning to believe he is right. Greg