Requirements Analysis: How to know what to do

A critical step in developing a GIS is to assess the needs and requirements for a potential system. This is a common problem in systems design of all kinds (information systems are just one example, and GIS is probably less formalized than most other applications).

It may be important to distinguish "system requirements" (choice of specific hardware and software) from the more conceptual "user needs assessment" that should precede and guide the system requirements.
One sequence to implement a GIS (simplified from Ventura and other sources):

Table 10-1: Sequence of steps to implement a GIS


0. Preliminaries Create awareness

Day 1: Agreement to begin implementation process

1. Needs Assessment

[Performed by interviews, surveys, reading old reports, etc.]

  • Data oriented sources, flows, transactions
  • Process oriented operations, transformations
  • Product oriented maps, graphs, tables, reports, decisions

    • 2. Requirements Analysis

    [Match needs to current technology]

  • Identify technical constraints (long-term)
  • System Design specify data models

    • 2.5 Results in an Implementation Plan

  • Generate Request for Proposals (RFP) (short term)
  • Evaluate responses to RFP
  • Select vendor [qualifications of vendors in next lecture]
  • Benchmark [or pilot project] measures performance of system

    • 3. Construction

  • Pilot Project/ Demonstration trial run of proposed design
  • Conversion digitize existing records; convert operations
  • Quality Control evaluate information in system

    • 4. Reevaluation (continual?)
    [return to start? (0 or 1)]

    Alternative Approaches

    (Assessing Needs, Analyzing Requirements)

    Seat of the Pants

    Many systems are designed in an opportunistic way, letting the technology rule.
    You can only do what the software permits you to do, after all.
    Vision also tends to be limited to replacement of current functions.

    In this way, the software and the implementation are like a dysfunctional relationship where no one asks any of the hard questions.

    A somewhat formalized model: Carol Freedman (& Dangermond) 1984

    attempt to develop generic data model for municipalities

    Matrix of tasks requiring different data sources (will this really apply across all municipalities? if it does, then there is little need for consultants like Carol Freedman... fortunately it didn't work, each case requires reexamination)

    Table 10-2: Generic Municipal Tasks (some of the 33)


    selected from (Dangermond and Freedman, 1984, p.13 Table 1)

    "Task Analysis"

    a stream in the Software Engineering literature
    Examine work currently performed, looking for a common step of basic "tasks" that could be rearranged, or served differently.

    One method to carry out Needs Assessment:

    Steve Ventura (U. Wisconsin - Madison) 1991
    Implementation of LIS in Local Government (no longer $10 from Wisc. State Cartographer)

    Distinguishes conceptual assessment, detailed assessement, system requirements

    Data oriented:

    what kind of data are required

    Process oriented:

    why is it used? how is it used?

    Product oriented:

    what kind of results do users expect?

    Put all three viewpoints together.


    Techniques:

  • Users Survey;
  • Interviews;
  • Review mandates and written evidence

    • Develop requirements:

  • technical (hardware, software, network, etc. [eventually a checklist of functions, like the obsolete list from 1988: Technology Exchange Working Group, Federal Interagency Coordianting Committee on Digital Cartography, became FGDC]);
  • logical (system design, data dictionary (such as the FGDC standards efforts, such as roads, "biological nomenclature"), entities and relationships);
  • institutional (membership of steering committee, hiring, reporting channels);
  • economic (Who pays? Who benefits? Who will maintain it?)

    • Concrete Examples of Implementation Projects

    FGDC Framework project (attempt to assure common content nationwide)
    Basis of this goes back to the Mapping Sciences Committee report 1990: Spatial Data Needs: the Future of the National Mapping Program, the 1993 report "Towards a Coordinated Spatial Data Infrastructure of the Nation" and later reports.

  • USDA Service Center Strategy: 600 counties/year (motivated by savings estimated at $168 million/year) report to FGDC (lost), Full text of report (.pdf)
  • Wisconsin Land Information Program; ( Statutes; Land Information defined) [warning, page errors?] Integration and Modernization Planning (each county has a plan - each year), following these instructions (.doc). Annual survey of counties, exhaustive matrix of data...
  • Example of one Wisconsin county: Ozaukee Land Information program; their plan; one reason they can do this: money!
  • North Carolina Coordinating Council Guidelines: A standards approach

    • GIS Implementation was a part of US national political agendas in the Clinton-Gore era, but Homeland Security means a shift... In fact, the military mapping establishment (NIMA now NGA) is moving into the mapping of 130 metropolitan areas. What is Geospatial Intelligence anyway?

    Index from Here: Resources for this lecture: GIS Marketplace (a whole lecture on its own)| Next Lecture | Schedule of Lectures | Labs and Due Dates | How to reach us
    Version of 1 December 2003