Apriori Implementation Team - Minutes
July 16, 1996 (1:00 - 2:30)

Attending: Mike Kearney, Mike Dettinger, Emily Corse, Jim Johnson, Lila 
Shapiro, Kristin Nelson

Handouts:
- none

Meeting Summary:
- NOTE TIME CHANGE!  Future meetings are planned for 1:30-3pm on 
Tuesdays, in Ira Winston's office at 3401 Walnut.  (Room has been reserved.)
- We will implement advice from Eleey, May, O'Donnell, and Winston to 
improve demo for Steering Committee.
- We will schedule another "practice" demo for ISC Support Providers and 
SAS Help Desk.
- Jim Johnson reported on progress with client data load.  We have 
requested initial load by August 5.  Change log will take longer, but 
will include nightly updates.
- Obtaining org code data presents certain challenges.
- We need to move forward on making Apriori operational at help desks.
- We need to investigate reporting functions and create necessary reports 
for operations.

Issues:
- Can Data Administration get us initial client data load by August 5?  
If not, will we need to push "go live" back until after fall crush?
- Do we need/ can we create a separate installation of Apriori for 
testing purposes?
- We must get access to org code data, which reside only in the 
warehouse, but the warehouse was not designed to support operational systems.
- Once we have org codes, we need to arrange for their translation to 
short, meaningful text description.

ToDo:
- MD & EC:  revamp demo.
- MD & EC:  schedule second practice demo w/ ISC & SAS staff.
- EC:  ask Rick Sabag for requirements and recommendations re: test system
- MK & KM:  work with Jeanne Curtis and Robin Beck on obtaining access to 
org data from warehouse.  We may need a resource who, given 50,000 SS#s 
from PennCard extract, could retrieve org codes from warehouse.

Meeting Details:
- We all agreed the "pre" demo went well.  Based on feedback from Eleey, 
May, O'Donnell, and Winston, MD & EC will revamp the demo to include a 
distributed staff scenario and a walk-in (rather than phone) intake.  We 
will add a segment that shows one or two sample canned reports.
- Mary Griffin has offered the ISC support staff as another guinea pig 
audience for practicing our revamped demo.  We can invite members of the 
SAS Help Desk as well.  The morning of 7/25 was offered, but won't work 
for EC.  For that demo, we will ask the audience to record questions 
during the demo and we'll try to respond at conclusion (or we'll gather 
questions and mail answers to group later--the beginnings of an Apriori 
FAQ!).
- Jim Johnson reported on client data:  Jim Choate and Bill Gilmore are 
writing specs for a generic extract, based on our data needs.  Fred 
Clothier, a consultant, will be hired to develop Oracle extract 
application (funding source unclear).  The initial load of approx. 50,000 
records should be straight-forward and could happen as soon as our 
requested August 5 deadline.  The change log presents some challenges, so 
add/ change/ delete will take longer.  Without the change log, we could 
possibly add new students (~2k records?) by reloading full set (~52k) or 
writing customer extract for new records only.
- JJ also discussed difficulty of obtaining org data, which does not 
reside in PennCard/People database, but only in the data warehouse.  The 
warehouse was not designed to support operational systems, so that 
joining PennCard/People to the warehouse is not routinely permitted.  JJ 
asked for help from KM & MK on arranging this access with Data 
Administration.  The worst-case scenario would require us to do a massive 
extract from the warehouse, based on the SS#s extracted from 
PennCard/People.  [In the long run, Penn would benefit from adding org 
code data to PennCard/People databases, but this will not happen soon.]  
In addition, once we have org codes, we need to arrange for their 
translation to short, meaningful text values.  We agreed that shouldn't 
be difficult, but must be planned.
- JJ would be interested in researching any data anomolies we discover, 
so we should create a mechanism for reporting these to him.
- In the context of adding bulk data to the production system, we 
discussed the possibility of needing a separate "test" installation of 
Apriori.  EC will ask Rick Sabag for requirements and recommendations.
- EC & MD need to investigate implementing Apriori at First Call and SAS 
help desk, at least as knowledge base.
- MK will investigate reporting functions.

Minutes recorded by E. Corse.  Please notify of any changes.