Moss L.T., Atre S. Business intelligence roadmap: The complete project lifecycle for decision-support applications

Подождите немного. Документ загружается.

7. Define preliminary

service-level agreements

Identify or revise the

expectations (outermost

acceptable limits) of the

business people in terms

of:

- Availability

- Security

- Response time

- Data cleanliness

- Ongoing support

8. Write the application

requirements document

Fill in the following

sections of the

application requirements

document:

- Technical

infrastructure

requirements

- Nontechnical

infrastrucure

requirements

- Reporting

requirements

- Ad hoc and

canned query

requirements

- Requirements for

source data,

including history

(include the high-

level logical data

model)

- Data-cleansing

requirements

- Security

requirements

- Preliminary SLAs

北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视仅供免费试用/版权原著所有

511/631

5. Data Analysis

1. Analyze the external

data sources

Identify the entities and

relationships from each

external data source

Merge the new entities

and relationships from

the external data sources

into the logical data

model

2. Refine the logical data

model

Fully attribute the logical

data model to include all

required source data

elements from internal as

well as external data

sources

Create new entities and

relationships where

needed to store the new

attributes (data

elements)

Analyze the layout of all

identified source files and

source databases, and

normalize overused data

elements, which are

explicitly redefined; for

example, a "redefines"

clause or an "occurs"

clause in a program

Analyze the content of all

identified source data

elements in the source

files and source

databases, and

normalize overused data

elements, which are

implicitly redefined; for

example, when a data

element is used for

multiple purposes

Create the data-specific

business meta data

components for all

attributes (data name,

definition, domain,

length, type, and so on)

北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视仅供免费试用/版权原著所有

512/631

3. Analyze the source data

quality

Apply business data

domain rules and

business data integrity

rules to find data

elements with invalid

domains, such as:

- Default values

- Missing values

- Cryptic values

- Contradicting

values (between

two dependent

data elements)

- Values that

violate the

business rules

- Missing primary

keys

- Duplicate primary

keys

Determine the severity of

the problem: how many

data elements have

invalid domains and how

many records are

affected; for example,

140 data elements out of

260 and 609,772 records

out of 2,439,087

Determine the criticality

of the problem (how the

dirty data will affect the

BI application if not

corrected)

北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视仅供免费试用/版权原著所有

513/631

4. Expand the enterprise

logical data model

Note

This activity is

typically

performed by

data

administration

behind the

scenes of the BI

project.

However, if the

core team

member who

plays the role of

the project data

administrator

must also play

the role of the

enterprise data

administrator,

the BI project

schedule may be

affected.

Merge the project-

specific logical data

model into the enterprise

logical data model

Identify data

discrepancies and

inconsistencies between

the logical data models

5. Resolve data

discrepancies

Discuss the discrepancies

and inconsistencies with

the BI project team, the

data owners, and the

business executives who

use the data in question

for their business

decisions

Adjust either the project-

specific logical data

model or the enterprise

logical data model, as

appropriate

Notify other project

teams that are affected

by the changes

If changes cannot be

implemented, document

the discrepancies and

inconsistencies as meta

data, and schedule a

北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视仅供免费试用/版权原著所有

514/631

time for the changes to

be implemented

6. Write the data-

cleansing specifications

Review the classification

of data elements: critical,

important, insignificant

Write data-cleansing

specifications for all

critical data elements

Write data-cleansing

specifications for selected

important data elements

(let the business

representative make the

selection)

Unless there is sufficient

time on the project and

the business

representative

specifically requests it,

do not write data-

cleansing specifications

for the insignificant data

elements

6. Application

Prototyping

Note

Prototyping is

an iterative

process, and

activities in

this step will

be repeated

multiple

times.

1. Analyze the access

requirements

Review the application

requirements document

with the subject matter

expert and the business

representative, and

together analyze:

- Report

requirements; for

example, how

many reports look

similar and can

therefore be

combined due to a

pattern in the

reports or use of

the same facts and

dimensions

- Query

requirements; for

example, do the

business people

want

parameterized

queries and if so,

what are the

北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视仅供免费试用/版权原著所有

515/631

parameterized

variables and who

will maintain the

query library

- Ad hoc

requirements; if

possible, try to

determine what

type of ad hoc

questions the

business people

may want to ask

based on what

types of questions

they currently ask

- Interface

requirements; for

example, a GUI

front end or a Web

portal

Communicate all your

findings to the database

administrator

Create a skill set matrix

for each business person

participating in the

prototyping activities:

- Indicate

computer skill level

as beginning,

advanced, expert

- Indicate

application

knowledge as

beginning,

advanced, expert

北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视仅供免费试用/版权原著所有

516/631

2. Determine the scope of

the prototype

Determine the objective

and the primary use of

the prototype

Decide which type of

prototype to build:

- Show-and-tell

- Mock-up

- Proof-of-concept

- Visual-design

- Demo

- Operational

Select a subset of

functions (reports,

queries, ETL, interface)

Select a subset of sample

data from the source files

and source databases

Create a change-control

document for managing

scope changes during the

prototype

Create an issues log for

tracking and resolving

issues during the

prototype

Determine the number of

prototype iterations

Determine the number of

prototype participants

(IT and business people)

Determine the time limits

for each prototype

iteration (time-box)

Estimate the cost and

benefit for each

prototype iteration

Determine the point of

diminishing returns for

the prototyping effort

北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视仅供免费试用/版权原著所有

517/631

3. Select tools for the

prototype

Review existing in-house

tools and find out who

uses them

Review the availability of

new reporting and

querying tools

Review existing or new

graphical tools

Review existing or new

report distribution tools

Review existing DBMS

options for the prototype

Select the platform on

which the prototype will

be developed

Select one of the

installed and tested

DBMSs

Select one or more

existing or new tools

Determine training needs

for the new tools

Schedule training

sessions as soon as

possible

4. Prepare the prototype

charter

Write the prototype

charter with information

about:

- Why you are

building the

prototype

(purpose)

- What type of

prototype you

selected (show-

and-tell, mock-up,

proof-of-concept,

visual-design,

demo, operational)

- Who will

participate (IT and

business people)

北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视仅供免费试用/版权原著所有

518/631

- What the rules

are (scope, time,

iterations)

- How you will

measure your

degree of success

5. Design the reports and

queries

Design the reports based

on existing reports or

mock-up report layouts

Design the queries based

on existing spreadsheets

or reports or on mock-up

samples

Design the interfaces:

GUI or Web front end

Create a physical data

model (database design)

for the prototype

database

Identify the data to be

used for the prototype:

either a representative

sample of source data or

new test data

Map sample source data

or new test data into the

prototype database

6. Build the prototype

Create the physical

prototype database

(tables, columns,

indices)

Create sample test data

(extract sample source

data or create new data)

Load the prototype

database with sample

source data or sample

test data

Write a selected subset

of reports

Write a selected subset

of queries

北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视仅供免费试用/版权原著所有

519/631

Write a selected subset

of interfaces or other

functions

Test reports, queries,

interfaces, or other

functions

Document any problems

with the tool

Document any issues

with the reports or

queries

Document any issues

with the interfaces or

other functions

Document any issues

with dirty source data

(do not waste time

resolving them unless

you are prototyping an

ETL function)

Using the prototype as a

yardstick, validate the

time and cost estimates

for the BI application

7. Demonstrate the

prototype

Review reports and

queries with the business

people

Review problems and

issues with the business

sponsor and the business

representative on the

core team

Review the project

requirements with the

subject matter expert

and the business

representative

Document requested

changes in the change-

control document

Analyze the impact of

requested changes in

terms of:

- Time

北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视仅供免费试用/版权原著所有

520/631