Debt Analysis Expert System
Vr.7 sof: spec.doc 1.3.4. Functional Requirements for the Enhanced Pilot of the KBS. This is a specification for a low budget enhanced pilot of the Knowledge Based System at West Midlands in the pre-litigation closed accounts section. Contents. --------- 1.3.4.1 Objectives. 1.3.4.2 Background. 1.3.4.3 Overview of the proposed system. 1.3.4.4 Data Extract Functionality. 1.3.4.5 PC based Functionality. 1.3.4.5.1 User Interface. 1.3.4.5.2 Reports. 1.3.4.6 Knowledge Refinement. 1.3.4.7 Data Backup. 1.3.4.8 Management Information. ----------------------------------- 1.3.4.1 Objectives. To provide specfication for the construction of a production PC system to enhance pre-litigation debt handling. 1.3.4.2 Background. The pilot project to demonstrate the technical feasibility has been completed. The pilot has demonstrated the value and feasibility of a KBMS knowledge base running under the OS/2 operating system on a PC. The system worked for a month analysing data downloaded from the ICL mainframe via floppy disk. A simple enhancement of the pilot system will require about 55 person days of effort to modify it into an acceptable production system. 1.3.4.3 Overview of the proposed system. A diskette will be delivered daily from computer services containing notice 34 data downloaded from the ICL mainframe. This will be loaded onto a dedicated PC by an automated process initiated by the users. Once the data is loaded the users will run the KBS to make decisions about each account for a particular extract date. Once the analysis has been run a report can be run showing the computers decisions for each account that has fallen into debt. 1.3.4.4 Data Extract Functionality. This will be the phase 2 data ie. the same as the pilot run (phase 1) but including court data. 1.3.4.5 PC Based Functionality. 1.3.4.5.1 User Interface. This will be an OS/2 version 2 presentation manager interface, ie. mouse and icon driven. Icons will start processes once clicked. LOAD DATA Start the loading of the download disk. RUN (ANALYSE THE CASES) This process will make judgements on all accounts for a particular extract date. REPORT This will start a report running showing the results of the run. 1.3.4.5.2 Reports List of Reports. ---------------- a) Decision Explanation. b) Decision List. c) Decision Analysis. d) Decision Summary. e) Value Analysis. Automated print of 'Decision List'. ----------------------------------- There will be five standard reports availble from the menu. The 'Decision Explanation' is the decision comparison report from the pilot. When the 'Decision Explanation' is printed there should automatically be a print of the 'Decison List' at the start of it. The 'Decision List' summarises the days cases and will be of used to supervisors. Variable time-span for all reports. ----------------------------------- As far as possible all reports should allow the user to specify the time span for report to act on. The user will type start and end dates. The report will then produce an analysis of all the data for the extract dates falling on or between the two dates. a) Decision Explanation. This gives a page report on each case analysed by the computer. b) Decision List. This report will be a list of cases and the computers evaluation of what course of action should be followed with respect to the customer on the basis of the information held on the mainframe eg. account history data. It should be printed automatically at the start of the daily decision report. ----------------------------------------------------------------------- start of Decision List================================== DECISION LIST ============== From Extract Date :- 10/4/93 To Extract Date :- 30/5/93 ACC KBS NO. TITLE INITIALS SURNAME DIST DECISION OTHER SUE W/O EX-DATE --- ----- -------- ------- ---- -------- ----- --- --- ------- 5252 MR O BROMLY 1 SUE 0 8 3 3/4/93 6256 MISS JJ GREZEBAD 3 OTHER 7 3 3 3/4/93 5252 MRS DIGBY 1 SUE 0 8 3 3/4/93 6256 MISS JJ JONES 3 OTHER 7 3 3 3/4/93 5252 MR O FRAZER 1 SUE 0 8 3 3/4/93 6256 MISS JJ GREZEBAD 3 OTHER 7 3 3 3/4/93 5252 MR O BLOGGS 1 SUE 0 8 3 3/4/93 6256 MISS JJ SMITH 3 WRITE-OFF 7 3 3 3/4/93 end of Decision List==================================== ----------------------------------------------------------------------- c) Decision Analysis. This report is decision by value. All the OTHER decisions are listed in order of ascending value, then similiarly for the sue decisions and the write offs. For the OTHER decisions the scores for sue and write off are also required. This is so the user can see what the computer would have decided if an OTHER consideration had not taken precedence. ----------------------------------------------------------------------- start of Decision Analysis============================== DECISION ANALYSIS ================== From Extract Date :- 10/4/93 To Extract Date :- 30/5/93 DECISION ACC NO. INITIALS SURNAME DISTRICT BALANCE SUE WRITE OFF -------- ------- -------- ------- -------- ------- --- --------- OTHER 12334444 R Pringle 1 2000 3 8 73767777 J Bromly 3 800 0 2 12334444 R Singh 1 2000 3 8 73767777 J Whiteney 3 55 1 2 SUE 12334444 R Humphrey 8 2000 3 8 73767777 J Bromly 3 800 0 2 12334444 R Singh 1 2000 3 8 73767777 J White 3 800 0 2 WRITE OFF 12334444 R George 3 660 3 8 73767777 J Bromly 3 800 0 2 12334444 R Singh 1 444 3 8 73767777 J White 3 800 0 2 end of Decision Analysis================================ ----------------------------------------------------------------------- d) Decision Summary. This report is by decision. The number of cases in each decision and the total balance outstanding are given. ----------------------------------------------------------------------- start of Decision Summary============================== DECISION SUMMARY ================== From Extract Date :- To Extract Date :- KBS Decision Count Total Balance Outstanding ------------ ----- ------------------------- Other 310 1076.43 Sue 200 4646.32 Write off 94 3333.03 -------------------------------------------------------- OVERALL 604 1032.78 end of Decision Summary================================= ----------------------------------------------------------------------- e) Value Analysis. This report is value by decision. The cases are listed in order of the size of the debt. Within this value banding the cases are summarised by decision. The number of cases in each decision are given and their total value. ----------------------------------------------------------------------- start of Value Analysis================================= VALUE ANALYSIS =============== From Extract Date :- To Extract Date :- BALANCE BAND KBS DECISION COUNT OUTSTANDING ---- ------------ ----- ----------- 50 - 100 OTHER 79 5750.41 SUE 129 9132.45 WRITE OFF 40 2748.16 248 17631.02 100 - 150 OTHER 34 750.55 SUE 78 6542.33 WRITE OFF 120 6548.66 232 5631.13 200 - 250 OTHER 79 5750.41 SUE 129 9132.45 WRITE OFF 40 2748.16 248 17631.02 ....... ETC. 1000-1050 OTHER 79 5750.41 SUE 129 9132.45 WRITE OFF 40 2748.16 248 17631.02 ----- -------- OVERALL TOTAL 1300 40,44.04 end of Value Analysis=================================== ----------------------------------------------------------------------- 1.3.4.6 Knowledge Refinement. At the start of the simple enhancement there would be some knowledge refinement necessary to make use of the extra extract data available under phase 2 as opposed to phase 1 (the pilot). Further refinement may be necessary in the light of cases looked at during production running. 1.3.4.7 Data Backup. In order to allow for the accumulation of one or two years data the PC will need to have a disk size of 500 meg. Data from the extract disks from the mainframe will be loaded onto the hard disk. If data does not have to be removed from the hard disk then there is an automatic backup by keeping the original download disks. The number required during one year would be about 52 * 5 = 260. This is the cheapest form of backup and has the advantage that it doesn't generate extra work and that it happens automatically as the data is loaded. The disks would have to be labelled with their extract date and stored in a locked, fire resistant cupboard. 1.3.4.8 Management Information. Some reports will be generated from the system on a monthly basis. Over and above this it is likely that other management information will be requested. It is not possible to define what management information will be required. Because of this all the data should be stored. The data is potentially valuable and should be kept readily available so that a variety of management information can be generated as and when required. Because this system could form an example to other regions it is important that all data about what decisions it is making are available, as it is running and also on a historical basis, preferably going back years. =======================================================================